The mathematician who developed much of the logic underlying computer design predicted, in 1949, that we faced “a decade or more of ruin and despair.” He forecast wholesale unemployment because automation, he felt, would abolish jobs on an unprecedented scale. Despite his expectations, the number of people gainfully occupied in civilian pursuits increased from 59 million in 1949 to 63 million in 1955.

In 1955, a parade of witnesses testified before the Congressional Subcommittee on Automation that intolerable unemployment was in prospect unless automation was used wisely and well. Since 1955, the number of Americans with jobs has increased from 63 million to 71 million—a record number. In addition, 4 million people now hold second jobs—an increase of almost 2 million.

During this period in which the number of civilian jobholders increased by 12 million and the number of jobs by 14 million—the period predicted to include “a decade or more of ruin and despair”—real wage rates and per capita income also increased. The average hourly income of factory workers in 1949 was $1.80 (measured in 1965 dollars). The average hourly income of factory workers now is $2.60 (exclusive of fringe benefits), a more than 40 percent increase. Since the typical workweek is essentially unchanged (39 hours in 1949, 41 hours now), this has meant a more than 40 percent rise in the weekly and annual income of the average factory worker. The typical nonfactory worker wage rate and annual income rose 35 percent in this same period. We have had a remarkable rise in the wage income of the average worker at the same time that the total number of people with jobs increased.

Why is automation alarming?

In the face of these data, why do some cry that doomsday is coming? What is it about automation that causes alarm? Why is it that workers asked about their attitude toward mechanization feel no threat, yet appear frightened when asked about their feelings toward automation?

The hallmarks of automation, to distinguish it from mechanization or automatic methods, are its sensing, feedback, and self-adjusting characteristics. Because it senses changing requirements and adjusts without human intervention, it presumably does away with the need for human attendants or human labor. This is very fearful indeed to those who depend upon jobs for their livelihood.

Fear of automation can be traced to four sources. One is based upon the assumption that there is a fixed amount of goods that buyers want. Any new method that enables us to turn out more goods per man-hour will, it is believed, enable us to turn out the fixed amount of goods and services with fewer men. If a man helped by an automatic machine can produce twice as many widgets per hour as he formerly did, then, presumably, only half as many hours of work will be available for each man to do. If workweeks are not shortened, only half as many jobs could, it is asserted, be provided in these circumstances. The president of the United States used this sort of logic when he said “that approximately 1.8 million persons holding jobs are replaced every year by machines.”

The second source of fear springs from the idea that automation or cybernation is something more than the latest stage in the long evolution of technology. Automated machines controlled by computers do not simply augment muscle power as previous machines did. They replace and outperform human intelligence. In the future, machines will not only run machines, they will repair machines, program production, run governments, and even rule men. Union leaders will collect no dues and business will have no customers because, supposedly, there will be no production workers required. Human beings, it is believed, will be made as obsolete by these machines as horses were by the tractor and the automobile.

The third source of fear is our greater awareness of the people displaced by automation than of the other unemployed, and a greater concern for these people. Among the more than 3 million unemployed are several thousand persons laid off because their skills are not usable by concerns installing automated processes to replace previously used technology. Presumably, because these thousands possess only obsolete skills, there are not job opportunities open to them. Others who are laid off or who are among the unemployed because they have voluntarily quit their jobs are less worrisome because their skills are not obsolete and they will have new jobs in a few weeks.

A fourth source of fear is the high incidence of joblessness among the unskilled. It is felt that the unskilled are unemployed because automated production reduces the demand for unskilled workers. Any increases in the demand for labor occurring because of automation are believed to be concentrated on highly skilled workers.
Is the alarm justified?

Automation causes displacement. Some people do become unemployed because of it, although most firms retrain and place employees in new jobs when eliminating old jobs. However, automation does not create unemployment. The number of jobless men is not greater than it would have been if no automation had occurred.

It may seem paradoxical to argue that automation causes displacement but does not cause unemployment. Many observers point to specific persons unemployed as a result of this phenomenon. They fail, however, to point to the unemployed who found jobs because of automation. They fail to recognize those who would have joined the jobless if new technology had not been developed. They fail to see that automation causes redeployment, not unemployment.

We may grant that automation differs from other kinds of technology. Yet, we should not blind ourselves to history to the point of saying it is completely new. Perhaps the earliest automated device was the pressure cooker, invented by Denis Papin in 1680. He originated a pressure control that is still one of the most widely used regulators. Despite this automated device, cooks are still extensively employed.

During the 18th century, several types of automatic regulators were applied to windmills. An automatic, card-programmed loom was devised by Jacquard over 150 years ago. An automatic flour mill was built in 1741. Eighteenth-century steam engines were controlled by governors, which had the sensing, feedback, and resetting characteristics that are the hallmarks of automation. Despite increasing automation in the last two centuries, employment has risen continually.

In terms of a very recent type of automation, the use of electronic-data-processing equipment, a US Department of Labor study of large firms that introduced such equipment concluded that:

Despite the reduction in labor requirements for the tasks performed by the computers, total employment of the offices as a whole rose. Over the four years from December 1953 to December 1957, total office employment at 17 of the offices studied increased an average of 7 percent. . . . In 6 of the 17 offices, the increase was greater than 15 percent; in 7, less; and in 4 there was a decrease. Although the immediate effect of electronic data processing suggests some retardation in the growth of office employment, particularly part-time work, the experience of some offices suggests the possibility of expanding employment in new areas of office activity to handle information which had previously been uneconomical to acquire.

This experience of increasing office employment despite reduced labor requirements per unit of output is a specific instance of what has been going on generally in our economy. From 1919 to 1962, man-hours required per unit of output in the American economy dropped by 67 percent, yet total number of jobs rose from 42 million to 68 million. The tripling of output per man-hour did not reduce the number of jobs by two-thirds, as those who believe in a fixed amount of work available would predict.

The primary effect of automation is not a reduction in the number of jobs available. Rather, it makes it possible for us to do many things that otherwise could not and would not be done. Automation enables us to earn larger incomes and lead fuller lives. It will, in the future, literally make it possible to travel to the moon. It saves lives through the aid it gives doctors. By controlling traffic signals in response to traffic flows and reducing traffic congestion, it adds hours to the free time of commuters every week. It helps scientists, with the aid of high-speed data processing, to develop new knowledge that otherwise would not be available in our lifetimes. We are increasing the scale of educational activities because mechanization, automation, cybernation, or whatever we choose to call our new technology makes it possible to do more than we could formerly. With the coming of automation, men are able to do more and have more. Both sublime and mundane activities are being enlarged.

How automation works

We know from experience that automation in the factory turns machine operators into machine tenders and maintainers. This has already occurred in the textile industry, to name one example. Upon walking into the loom room of a modern mill, one has the first impression of a vast space filled with busy machinery and no people in sight. (Yet employment in textile mills totals nearly 900,000 workers.) Controls on individual machines enable one man to supervise a dozen or more looms.

The effect of automation has been to increase the relative number of maintenance men, engineers, office employees, production control specialists, and other nonmachine operators that are required. (These are the people the US Census Bureau calls nonproduction workers.) This is simply a continuation of a trend that has been going on for decades. In 1899, only 7 percent of the manufacturing industry labor force consisted of persons other than production workers. Today 26 percent of manufacturing employees are nonproduction or indirect workers. Since 1939, production-worker employment in manufacturing has increased 65 percent, while that of other workers has increased by over 160 percent.

In addition to changing the balance among occupations in a given industry such as manufacturing, technological progress is also changing the balance among industries. Only a century ago, 50 out of every 100 workers toiled on farms producing the nation’s supply of food and fiber. Only two to three out of every 100 workers were producing educational, medical, recreational, and other services that contribute to a richer, fuller, healthier life. Today, the number of workers in these life-enriching occupations is relatively five times as great. Those toiling on farms have been reduced relatively to one-sixth their former number. They now direct machines instead of using animal power and their own muscles. The quality of life has been improved and brute toil has been reduced because technology has increased our incomes to the point where we can afford these services and these machines.

Those who are concerned about unemployment should welcome rather than fear automation. If it were not for the technical advances of the past decade, unemployment, at present wage levels, would be above the astronomical levels of the early 1930s. Alternatively, if real wage rates were at levels consistent with full employment using the same technology as that available a decade ago, wage rates would be lower by about $8 a week (or 20 cents an hour) than they are now.

Technological change has created more jobs than it has destroyed. The number of civilians at work in 1965 is 8 million higher than a decade ago. A number of forces including advances in technology created nearly 60 million additional jobs during the past decade. More than 50 million jobs were destroyed, however, by various causes (primarily by the upward movement of wage rates). The nearly 60 million new jobs created less the more than 50 million jobs destroyed left a net gain of 8 million positions.

Does automation kill jobs?

If automation creates jobs and raises the productivity of those with jobs sufficiently to make it possible for them to earn more, then why is unemployment among teenagers now at the 14 percent level, four times the unemployment rate of adults? Are we faced with a situation in which jobs for the unskilled and the inexperienced are being wiped out by automation?

In this case, the primary cause of unemployment is the overpricing of many jobs that would normally be filled by inexperienced, unskilled new entrants to the labor force. The unemployment among teenagers is a consequence of arbitrarily determined wage rates for certain groups of jobs, which have caused a contraction of employment opportunities for the unskilled, inexperienced worker.

There is a growing concentration of unemployment among unskilled workers not only because of the high minimum wage rates for newly hired workers set by union-employer agreements, but because of the successive increases in statutory minimum wage rates by congressional amendment of the Fair Labor Standards Act. In 1949, the minimum-wage rate set by federal statute was 40 cents per hour. At this time, average earnings of manufacturing industry employees were $1.38 an hour. The minimum wage has been increased several times since 1949, reaching $1.25 in September 1963. At that time, average earnings of manufacturing industry employees were $2.47 per hour. Thus, compared with the hourly wage of workers in manufacturing, the minimum wage rose from 29 percent in 1949 to 51 percent in 1963.

It is hardly surprising that unemployment among the unskilled increased with this rapid rise in the minimum wage. In absolute terms, the statutory minimum wage has more than tripled since 1949. In relative terms, it has risen to 176 percent of what it was in 1949.

Increasing the price of unskilled workers so greatly relative to that of skilled workers unduly penalizes the hiring of the unskilled. It is fortunate that the proportion of the work force that is unskilled has been diminishing. Otherwise, the unemployment problem would be more severe than it is, given the increases in minimum-wage rates that have occurred.

The doom-criers who are alarmed about automation say that “a permanently depressed class is developing in the United States.” If there is such a class, it is caused by legislation such as the Fair Labor Standards Act, not by automation. However, the data on income received by the poorest 20 percent of the population do not indicate that those people are becoming worse off. From 1949 to 1962, average family income of the poorest 20 percent of the population rose by 60 percent in current dollars, or by 28 percent measured in constant dollars.

If no technological change had occurred in the past decade, the number of jobs available could have grown as it has, from 63 million to 71 million, only at the price of restricting increases in wage rates. Wage rates could have been increased by only 30 cents per hour or $600 per year instead of $1,000 a year. With this restricted wage increase and automation, the number of jobs would have grown to 91 million instead of 71 million. In effect, technological change created 20 million jobs in the past decade.

Instead of castigating automation for causing unemployment, we should be inviting more automation to help solve the present unemployment problem. The overpricing of labor in industries such as coal-mining and the setting of high minimum-wage rates by statute for unskilled labor have caused unemployment because many are not productive enough to be employed at these wage rates. With more technological advances, productivity would be increased. The workers presently priced out of the market would be employable if their productivity were increased, and it would be by technical progress.

Automation should be welcomed as the means of alleviating poverty and undoing the damage done by bad wage laws and improper union-employer agreements. It should not be feared as a job destroyer. It is a job creator.

Yale Brozen was professor of business economics from 1957 to 1987 at Chicago Booth. He died in 1998.

More from Chicago Booth Review

More from Chicago Booth

Your Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.