NOTHING HAS done more to encourage U.S. companies to adopt Japanese management methods than the Toyota-General Motors joint venture in California -- New United Motors Manufacturing Inc. (NUMMI). In just two years, the Fremont assembly plant went from an "interesting experiment" to the success story in U.S. automobile manufacturing.

NUMMI, which makes the Toyota Corolla and Chevrolet Nova, has made U.S. managers sit up and take notice for several reasons:

The automaker has achieved a massive increase in labor productivity. By 1985, GM could claim that NUMMI required only 14 hours of direct labor to assemble a Nova. By contrast, it had taken 22 hours to produce the J car in the same plant before GM closed it in 1982.

The start-up used unfamiliar equipment, workforce methods and management, yet quality climbed quickly and has remained outstanding.

NUMMI achieved these feats in a traditional assembly plant, where the technology is not particularly advanced. Furthermore, most of the workers are the same ones who previously had a reputation for wildcat strikes, a lack of concern for quality and major problems with drugs, alcohol and absenteeism.

Local and national union leaders are some of NUMMI's biggest boosters.

For U.S. corporations, spurred by the recession of the early 1980s, declining profit margins and growing foreign competition, the lesson is clear: You can take an ordinary factory with a unionized work-force and, by changing the management and contract, achieve productivity and quality to compete with the Japanese. What is far less clear is that this is often accomplished at great cost to a humane work environment.

By March of this year, the "team concept" -- as the NUMMI model is usually known -- was in use or planned for at least 17 GM assembly plants -- including the Saturn project -- as well as various Chrysler and Ford facilities and all of the wholly or partially Japanese-owned auto plants in the United States. The United Auto Workers has endorsed the idea in national contracts with the Big Three. And the team concept is growing in other industries, from corn syrup to insurance, electronics, chemicals and paper. Among those experimenting with the idea are AT&T, General Electric, Proctor and Gamble, Xerox and Honeywell.

Putting On the Squeeze

At plants such as NUMMI, a team usually consists of four to 15 workers plus a team leader, who has no fixed production job. Over them is a supervisor (called "group leader" or "adviser" now rather than "foreman") who supervises two to four teams. Yet despite the terminology, in actual operation there is no more teamwork, in the sense of workers helping each other, than in a traditional auto plant.

In addition to "teams," managers at NUMMI and other Japanese-run plants have implemented a total approach to production that includes "just-in-time" inventory control, technology designed to minimize indirect labor, extensive use of outside contracting and systematic speed-up. This system is sometimes called "synchronous manufacturing." It might more accurately be described as "management by stress" -- which also conveys the daily experience of those who perform the labor in such plants.

To identify both weak and strong points, the system -- including its human elements -- operates in a state of permanent stress. The weak points break down, indicating where additional resources are needed. Just as important, points that never break down are assumed to waste resources.

The andon board illustrates the principle. Hung over the NUMMI assembly line, it displays the status of the operation with a separate rectangular area for each workstation. When the worker falls behind or needs help, he or she pulls a cord and the rectangular area lights up. If the light remains on for a set period (such as one minute), the line stops. {See box}

In a traditional operation, management would consider no lights to be the desired situation. Individual managers would protect that status and themselves with excess workers. In management by stress, however, unlit warning lights signal inefficiency. If, for example, the line is speeded up, the lights will quickly reveal the weak points. At some points, the line will stop, and management can focus on redesigning those jobs. The ideal is for the system to run with all stations oscillating between lights on and lights off.

Stressing the system can be accomplished by increasing line speed, cutting the number of people or machines, or giving workers more tasks. Similarly, a line can be "balanced" by decreasing resources or increasing the workload at positions that run smoothly. Once problems have been corrected, the system can be further stressed and then balanced again.

"Team-based systems," Business Week editorializes, "are alternatives to the scientific management system, long used in Detroit, which treats employees as mere hands who must be told every move to make." This is part of the fantasy being constructed around the team concept. In fact, the tendency is the opposite: To specify every move that a worker makes in much greater detail than ever before.

In team-concept plants, as in traditional plants, workers have little control over basic job design. Management chooses the processes, basic production layout and technologies. While jobs are designed by "teams," the members of those teams in the pre-production phase are engineers, supervisors, and management-selected group and team leaders. They "chart" the jobs -- break them down to separate acts -- and then shift tasks so that the amount of work individual members have to do is more or less equal. The result is a detailed written specification.

Workers do get to participate some in fine-tuning the system. It is this phase that inspires most of the rhapsodies about team concept. But once full production is reached and bugs eliminated, the jobs are "standardized" -- done in precisely the same way every time. The team member is told exactly how many steps to take and what the left hand should be doing while the right hand is picking up the wrench. Workers are not allowed to vary their techniques to accommodate individual variations such as height or agility. They may not work faster for a short time to try to create some breathing space. (The jobs are so "loaded" that this is usually not possible anyway.) If they do discover a method that makes the job easier, they must share it with the supervisor, who will then take advantage of the time saved by adding another task to the job.

There is always room for kaizen, or continuous improvement. In his book, "Toyota Production System: Practical Approach to Production Management," Yasuhiro Monden gives an example: Management wants to reallocate jobs on a team because five workers are working every second, but the sixth has 45 seconds of idle time. This waiting time, says Monden, should not be distributed equally among the six workers. "If it were, it would be simply hidden again . . . . Instead a return to step 1 is necessary to see if there are further improvements that can be made in the line to eliminate the fractional operations" for worker number six.

Once the line is up to speed, each worker can barely keep up with the specified job, let alone help someone else. In fact, a worker who can shave a few seconds off a task shouldn't help fellow workers: It is better to stand idle to reveal the existence of free time, which can then be used on a regular task. At NUMMI, when slow car sales necessitated a cut in the line speed, jobs were rebalanced so that employees were still working as nearly as possible to 60 seconds out of a minute. Because management had promised that there would be no layoffs, those who became extra were assigned to observe their fellow workers and kaizen their jobs even further.

Supervisors know the production process because they regularly work the line, a practice forbidden by traditional auto contracts. The team leaders, who are union members but are often effectively incorporated into management, work all the jobs in their teams from time to time and are also expected to help "improve" those jobs. At NUMMI, some 900 workers have taken the 30-hour (unpaid) training course to qualify for a team-leader position.

Another essential ingredient is "just-in-time" (JIT) production, a "demand-pull" approach whereby parts are made only when the next operation needs them. No buffer stocks are kept on hand. JIT saves interest costs on capital tied up in inventory and reduces the expense of warehousing. It also saves on quality control since faulty parts are detected quickly. As with the andon board, management by stress uses JIT to tighten departments that never have a problem. JIT rejects the traditional "just-in-case" rationale for inventory -- protecting one department from problems in another. Instead, by removing the cushions, JIT reveals inefficiencies.

The Human Machine

Automation is never adopted for its own sake. A machine that replaced a production worker but required an additional electrician to service it would be rejected. Furthermore, automation must increase management flexibility, not decrease it. Monden warns that "Even if the introduction of an automatic machine reduces manpower by 0.9 persons, it cannot actually reduce the number of workers on the line unless the remaining 0.1 person . . . can be eliminated." Further, "the introduction of {automation} may actually eliminate the ability to reduce the number of workers -- a matter of some concern, since it is always essential to reduce the work-force, especially when demand decreases."

Yet ironically management by stress makes it easier to introduce more advanced or larger-scale technologies. Since workers must do a job in a machine-like way, management uses them as a prototype for automation; and the ability to assign workers flexibly to new duties and to force "cooperation" bypasses many of the usual obstacles to introducing automation.

If JIT forbids producing in advance and idle time cannot be tolerated, production must be organized so that jobs can be shifted and adjusted easily when demand rises or falls. The ease and speed with which management-by-stress plants respond to the ups and downs of the market contribute greatly to their high productivity.

Such management flexibility requires that workers be able to perform many jobs. The common name for this is "multi-skilling," but the term is misleading. The essence of multi-skilling is actually the lack of obstacles -- such as contract provisions, job classifications or tradition -- to management's desire to reassign jobs. Thus hiring practices put little emphasis on applicants' skills. Production jobs that have been carefully broken down into simple actions do not require intensive training or specialized knowledge, only manual dexterity, physical stamina and the ability to follow instructions precisely. Once hired, workers do not learn new marketable skills.

Down on the Line

Team-concept advocates claim that the dignity of the individual is central to their theory. Toyota managers call it the "respect-for-human" system, and there is superficial evidence of that respect. Visitors to NUMMI are struck by the plant's atmosphere. Workers are addressed using courteous language; the factory is clean and well lit and seems like a nice place to work.

But the operation of the plant indicates a very narrow notion of humanity: Fulfillment is achieved only by striving for management's goals. A manual used at the Michigan Mazda plant warns, "If you are standing in front of your machine doing nothing, you yourself are not gaining respect as a human being."

Mike Parker and Jane Slaughter are co-authors of "Choosing Sides: Unions and the Team Concept." A longer version of this article appears in the current issue of Technology Review.