TMG Home
Michael Maccoby, President
Richard Margolies, Vice President
Nora Maccoby, Vice President
Sandylee Maccoby
Speaking Engagements
Contact Us

Is There a Best Way to Build a Car?

by Michael Maccoby

Printed in: Harvard Business Review, November-December 1997, pp 161-170.

Books in Review

Farewell to the Factory: Auto Workers in the Late Twentieth Century
Ruth Milkman
Buy it Now Berkeley California: University of California Press, 1997 Just Another Car Factory? Lean Production and Its Discontents
James Rinehart, Christopher Huxley, and David Robertson
Buy it Now Ithaca, New York: Cornell University Press, 1997

During the past two decades, a profound transformation in automobile production has resulted in better quality and lower costs. The new system that has delivered these results, lean production, has become the competitive standard for companies worldwide. But a debate continues over how the system treats workers. This debate centers on not only economics and engineering but also human values and ideology.

Ever since 1972, when workers at General Motors’ assembly plant in Lordstown, Ohio walked out to protest a speedup of the production line, U.S. automakers have worried about the quality of life in the workplace.Social scientists, linking employees' alienation with poor product quality and high costs, argued that an increasingly educated workforce was demanding both more autonomy and more intellectually challenging work. In cooperation with the United Automobile Workers union (UAW), manufacturers tried to involve workers in problem solving and decisions about how to perform their jobs. Because the automakers didn't change the production system, however, they had limited success.

In the 1980s, lean production arrived in the United States from Japan - a system that explicitly relied on workers' initiative in order to reach its goals on quality and productivity. As evangelized in The Machine That Changed the World, a report sponsored by the Massachusetts Institute of Technology, lean production would give employees responsibility for maintaining quality on the line and for designing work tasks. Through extensive training and job rotation, workers would develop a wide range of skills. The new system, the authors claimed, would not only reduced defects, inventories, and costs but also humanized work by giving employees more challenge and responsibility, (although at the same time possibly creating more stress.) The authors predicted “a creative tension involved in solving complex problems" so that the gap between manual and knowledge workers is narrowed.

The MIT report, combined with the success of Japanese factories in other countries, persuaded U.S. and European auto executives that lean production was the best way to build cars. Most U.S. auto factories have adopted at least some aspects of lean production, and according to John Paul MacDuffie of the University of Pennsylvania's Wharton School, one-fourth of U.S. plants have tried to adopt the entire system.

As the authors of two new books argue, however, the results have not matched the rosy picture painted by MIT group. In Farewell to the Factory, Ruth Milkman, a professor of sociology at the University of California at Los Angeles, examines a GM factory at Linden, New Jersey, that had adopted lean production in 1986. In Just Another Car Factory? James Rinehart and Christopher Huxley, sociology professors at the University of Western Ontario and Trent University, respectively, join David Robertson from the Canadian Automobile Workers union to report on a GM-Suzuki joint venture called CAMI (from the Japanese word for "deity"); the plant, located at Ingersoll, Ontario, began production in 1990. The two books show how each plant failed to make good on promises of worker participation.

A New Form of Taylorism

Despite company slogans calling for involvement and empowerment, the authors report, workers at the two plants experienced little that was different from traditional Taylorist mass production on an assembly line - with its narrow, standardized, monotonous tasks repeated every 60 to 90 seconds. Workers at Linden were told they could pull a cord and stop the line to prevent defective products from proceeding, but those who did so were criticized by foremen and eventually stopped trying. Workers at Cami expressed similar frustrations, saying that the plant had turned out to be little different from traditional factories. They decided management was exploiting them with lower pay and stressful working conditions, and they went on strike a few years after the plant opened. They expressed their dismay by bearing banners on which the company slogans of ‘open communication, empowerment, kaizen, and team spirit ” were crossed out and replaced by union ideals of ‘dignity, respect, fairness, and solidarity.’

The two books do present slightly different explanations for why lean production proved to be so disappointing. Milkman, whose book describes the experiences of workers under both the traditional and the lean systems (and includes a chapter on workers who opted for a buyout instead of remaining with the plant after the conversion), blames managers at General Motors for failing to gain the commitment of frontline supervisors to the new approach. The workers were eager to try out the new system, but foremen were used to the old dictatorial ways and were themselves under constant pressure from superiors to maintain a high volume - even while workers were adjusting to the new system. Rinehart, Huxley, and Robertson, by contrast, see an inherent problem in the system itself as it played out in a new plant: many managers wanted to give employees some discretion, they argue, but that desire "collided with a production system oriented to maximum output with minimum workers."

Both books agree that workers do not like automobile assembly. Even when supervisors have allowed teamwork and opportunities for problem solving and participation (all of which workers appreciate), assembly work remains monotonous and stressful. As a result of kaizen, - or continuous improvement - workers, rather than industrial engineers, can modify jobs. But the work remains much the same, and managers still make the final decisions. The jobs most prized by workers are the ones which allow the most autonomy, ranging from high paying jobs such as programming robots to lower-paying janitorial work. As the books point out, what really attracts people to these factories is not the work itself but the relatively high wages gained by a strong union.

These books' authors make a point of addressing the MIT group, but the two sides seem to sail by each other-one focused on working conditions, the other on production efficiency. Milkman and Rinehart, Huxley and Robertson criticize lean production for pretending to empower workers while actually increasing managerial control. At Cami, participation was a means of having workers themselves eliminate jobs - a managerial tool for a new version of Taylorism in which employees help design the system that oppresses them. At Linden, the pretense of participation soon disappeared as foremen were driven by demands to cut costs and increase output. At both plants, managers used whatever participation they allowed merely to harvest workers’ knowledge and reduce conflict.

While the MIT group paid little attention to the realities of workers' attitudes and values, the inevitability of conflict, or the role of unions, the authors of the two books take little note of the economics of the industry or the demands of competition, and they offer no alternatives to lean production for fostering participation and achieving gains for both workers and companies. Although the GM’s and Suzuki’s managers come across as being more traditional than they claimed to be, both plants adopted aspects of lean production and improved their results substantially.

Success Stories

How humanized can manufacturers make assembly work, given the stress of tough repetitive activity? One solution is to do away with the assembly line altogether, but that may be impractical. Volvo’s factory in Uddevalla, Sweden, experimented in the 1980s with teams that assembled cars at their own pace and allowed workers to rotate from the assembly area to the highly automated parts-delivery system. This new approach sought efficiency through employee initiative and learning, but it did not achieve productivity targets. Volvo closed the plant in 1993 when it was negotiating a merger with Renault, an automaker that favored learn manufacturing. The MIT report dismissed the Uddevalla plant as a romantic attempt to return to the craft age, although Swedish academics maintain that the factory was on a steep learning curve that eventually would have made it competitive in terms of productivity and superior in terms of employee satisfaction.

Assembly line work will probably always be undesirable, but employees find such work more engaging when they can make improvements based on human as well as economic criteria. Any chance of that type of participation happening at Linden and Cami slopped away as distrust between union and managemetn grew. Yet two of the best examples of cooperation, high morale, and successful production are actually in the GM system: NUMMI (New United Motor Manufacturing, Inc.), a joint venture with Toyota established in 1984 in Fremont, California; and the autonomous Saturn division in Spring Hill, Tennessee.

The NUMMI venture took over a GM plant that had been shut down because of worker unrest and low productivity. Despite inheriting a combative local union and workforce, Toyota's manager transformed the plant into a prime example of lean production - a model closer to the MIT ideal than any other unionized factory in North America, comparable in productivity to Toyota's Japanese plants and to the best nonunion transplants. Although Saturn has adopted some aspects of learn production, it organizes work around autonomous teams rather than following Toyota's leadership and standardization across teams. Still, Saturn is worth considering because it exemplifies one extreme on the spectrum of partnerships between union and management, with local union officials sharing management roles.

Neither NUMMI nor Saturn has been copied within GM. Saturn, although profitable, has not equaled the productivity of the best lean-production plants because its use of autonomous teams keeps it from standardizing process improvements. Perhaps more important, both corporate GM and the union's national leadership distrust Saturn's autonomy. GM wants to fold Saturn into corporate wide systems and expand the car's production into more traditional plants. The union fears that the local is becoming too focused on Saturn's success and unwilling to participate in efforts to benefit members at other GM locations.

GM's failure to learn from NUMMI is hard to explain. Both productivity and quality are at high levels. Indeed, the automaker sent managers and engineers to work at NUMMI in the hope that they would carry successful practices back to their own plants. But those factories have seen little improvement. Why? The answer will not come from those who believe that there is a single best way to build a car. Lean production is not merely a set of processes and procedures. It is a broad system that engages its participants in their work with an ideology that appeals to their hearts and minds. As Haruo Shimada of Japan's Keio University has said, the Toyota system amounts to humanware - combining management, participation, incentives, and technology strategy to optimize productivity, motivation, and the development of people at work.

Toyota's ideology manifests itself at NUMMI in many ways. While engineers at Linden automated jobs in order to cut costs, NUMMI has robots that are designed to make people's jobs easier - for example, robots that do heavy lifting. The idea is that technology should be a tool and that workers should add wisdom to the machines. Moreover, the automaker emphasizes the exercise of leadership at every level to create trust and facilitate learning - a point that U.S. manufacturers have neglected in focusing on how Toyota eliminates inventory, defects, and waste. At NUMMI, workers who teach and help others are promoted to team leadership roles. At Toyota's Japanese factories, foremen are even given a budget to entertain workers periodically, buy them a beer, and strengthen relationships within teams. The Toyota leader aims to treat employees with respect.

In "Time and Motion Regained" (HBR January-February 1993), Paul S. Adler, a sociologist who interviewed employees at NUMMI writes that workers are made to feel important; their views carry weight. Contrast this to Linden, where foremen yelled at workers when they tried to prevent defects by stopping the line. Toyota's managers have also developed a cooperative, problem-solving relationship with the same local union that fought GM. Although some NUMMI workers complain about the fast pace and about group pressure within teams, they still participate in on-going learning and continous improvement. At both NUMMI and Saturn, teams eliminate jobs, but not at the cost of laying off workers. In fact, there have been no layoffs or speedups at either plant.

Different paths to participation

Is this emphasis on leadership particular to Japan? The managerial ideal in cultures based on Confucian thinking is, to some degree, the benevolent but demanding teacher/leader who creates strong bonds of mutual obligation. But not all Japanese companies are alike. Honda's approach to lean production emphasizes creative conflict, whereas Toyota's approach stresses benevolent hierarchy. And Toyota's leadership ideology is rooted in a rural feudal culture, whereas Honda's team ideology emerged in urban Tokyo. The Toyota worker was typically a farm boy brought by his parents to the factory and handed over to management in loco parentis.; in contrast the typical Honda worker was a member of a motorcycle gang who brought a rough egalitarianism to a company that was itself rebelling against the Japanese establishment. (The government had disapproved of Honda’s entering an already crowded market.) Company lore recounts how the engineer who stood up to and disagreed with founder Soichiro Honda eventually became president.

Whereas Toyota and NUMMI attack problems with established teams and standardized methods, Honda boasts of spontaneous solutions. Everyone with relevant knowledge gets together to solve, say, a quality problem and then the group disbands. Honda breaks down boundaries between functions by requiring designers to spend time in production and engineers to sell cars. At the Marysville, Ohio plant, which is not unionized, managers and engineers regularly work on the line for up to an hour each day. Honda’s cross-functional teams that design new cars are not hierarchies but "heterarchies" in which leadership shifts according to who has the appropriate knowledge.

Toyota and Honda, in their unique ways, develop and align competencies, both technical and interpersonal, with the goals of lean production. In both cases, corporate ideology reinforces the system. Few U.S. companies understand or are able to create such effective systems - systems that integrate and shape employees' values and competencies with a strong ideology. Non-union Hewlett-Packard and unionized Southwest Airlines are notable exceptions, and so is Saturn. These companies have a team-oriented egalitarian ideology that is closer to Honda than to Toyota -- whose approach to leadership does not come naturally to Americans. Americans value autonomy above all else. Milkman’s interviews with the Linden workers reinforce what other studies have shown: People in the United States want to be their own boss. But, if they can't be, they want as much autonomy as possible. Americans may recognize they need leadership, but they would rather do the leading themselves.

A culture that values autonomy does not produce many leaders. Most business professionals in the United States become managers not because they want to be leaders, but because management is a chore they are required to perform in order to be promoted. There are, of course, remarkable coaches who develop people, but they are the exceptions. Most Americans get ahead in business not because they are good leaders, but because they get results and get along with the boss. Despite efforts by human resource departments to teach "people skills", the ideal business leader in the United States has been something like "The One Minute Manager" who delegates authority, makes brief appearances to reward or reprove, and defends his or her turf against threats to autonomy.

For example, both Toyota and Honda have created effective cross functional teams that use input to design cars for ease of production. Toyota does this with strong leadership and Honda with team competence. But when U.S. automakers have tried to adopt these approaches, they have floundered as managers of design or engineering have asserted their autonomy and defended their turf.

Yet as NUMMI demonstrates, U.S workers do respond to leaders who teach useful skills and business perspectives, communicate market and process information and reasons behind decisions, and facilitate problem-solving sessions. If they can’t be autonomous, employees at least want a say in how their work is done. In different ways, both NUMMI with its humanized hierarchy and Saturn with its egalitarian teams satisfy this need.

In addition, the degree of cooperation between union and management can vary in successful plants. At Saturn, union and company representatives manage the plant as partners. At NUMMI, the union representatives participate in planning and problem solving, but management makes the decisions. Toyota's managers at NUMMI respect the union’s legitimate role as representing its members ’ rights and concerns. The local UAW, for example, guards against increased job stress and negotiates improvements in working conditions and ergonomics. Indeed, when Toyota took over NUMMI, the number of worker grievances dropped, and the number of ideas from workers increased.

I once expressed my amazement to a foreman in Toyota City, Japan, that so many ideas were generated by workers - on the average, more than one a week, of which more than 80 percent were adopted. “You in the West have a different view about ideas then we do,” he said. “You are happy when workers are satisfied and there are no complaints. For us, every complaint is an opportunity for improvement. We encourage people to turn their complaints into ideas.”

A union’s ability to cooperate with management depends on management's accepting the union as a legitimate representative of the workers and on its willingness to involve employees in the decision-making process. When management becomes autocratic, as it is at Linden, a union will encourage resistance. Most workers look to the union to help expand their participation, their opportunity to have their voice heard directly rather than through a representative. But they also want protection from management taking advantage of such participation. As Saturn and NUMMI have discovered, even when management allows participation, a minority of employees will distrust union leaders who cooperate too closely with management. Indeed, workers in the United States distrust union leaders as much as they distrust anyone - in government or in business - who can place limits their freedoms. From their union representative, they want, first, a good business agent (“Show me the money”) and second, a tough, smart defense attorney who can protect them if need be. (And labor law tends to reinforces that approach.)

To justify expanding its role to include cooperation with management, a union must convince both its members and management that such a role furthers their respective interests. Members must believe that the union can secure participation for them with protections and fair process. The Communications Workers of America, for example, helped bring about greater participation in its industry by taking into account both employee and company needs. The union’s leaders found that market pressures and process reengineering were altering members' working conditions too rapidly to be covered under the usual three-year contracts, but they also recognized that the union's members shared a strong interest in the success of the companies that employed them. At AT&T, the result of these insights was an educational program that equipped teams to shape and evaluate work policies according to three criteria: profitability, customer satisfaction, and employee well-being. A planning council of management and union officials supports and coordinates changes recommended by teams, giving the union a role somewhere in between the UAW's comanagement approach at Saturn and the more easily sustainable watchdog, problem-solving position at NUMMI.

To develop effective and enduring union participation, both managers and workers need to be educated about business and union perspectives, and they need a meaningful ideology that explains the logic and value of their work. Workers and union representatives need to understand competition and financial requirements. Managers need to understand that a union is a democratic organization that functions as both a service business and movement for social justice. Both workers and managers need to understand what determines the value of their product to customers. Saturn, for example, continually educates its workers in business topics and problem solving, while NUMMI teaches its employees about the Toyota production system and provides market data on quality and internal data on efficiency. Linden, by contrast, introduced lean production with 80 hours of training, but that training focused mainly on the psychology of motivation and stress reduction

That neither Linden nor CAMI realized its promise does not invalidate lean manufacturing. As the two books efficiently show, however, the MIT's group's description of this system is misleading in that the group underplays its human ad social aspects, as well as the central role of ideology. Although it is unrealistic to expect assembly plants to become democracies, plants function best when workers understand the business and participate in continuously improving it and their own working conditions. A collection of best practices does not make a successful production system.

[an error occurred while processing this directive]

This page was last updated Tuesday, 10-Nov-1998 17:56:50 EST.