Meeting New Challenges for U.S. Industry
Toward a Learning Economy
More Americans now work in physicians’ offices than in auto plants. Roughly as many work in retailing as in all of manufacturing. The service sector now encompasses three-quarters of U.S. jobs, and the share will only grow. However, productivity is growing much more slowly in services than in manufacturing, wages in services lag those in manufacturing, and income inequality in services is much greater. Unless the United States shifts its focus to strengthening the service sector, the nation’s productivity, wages, and standard of living will wither in the 21st century.
Since the Civil War, manufacturing has served as the mainspring of U.S. economic growth. The manufacturing economy did not deliver sustained prosperity, however, until corporate organization and public policy were adapted to support mass production and achieve economies of scale. The final pieces of the puzzle were put in place after the Great Depression that caused wages and purchasing power to rise along with the newfound capacity to produce. Unemployment insurance, social security, and the minimum wage ensured that the jobless, the retired, and working poor would do their part to stoke the economic engine. Most critical to sustaining mass consumption, unions won middle-class pay for their members, with spillover effects pulling up the wages of many managers and nonunion workers.
Productivity and wages rose in tandem until the 1970s, when foreign competitors began to challenge U.S. manufacturers. Since that time, the service sector has continued to expand. In the future, manufacturing, although still important, will be too small to drive the U.S. economy. Services are the new driver. The people who fill service jobs may not wear blue collars, but they are the counterparts of those on the factory floors of the industrial age. How they fare in the coming decades will determine whether the fruits of the information age benefit all Americans or only those at the top of the income distribution. And how productively they work will determine whether U.S. economic performance improves at a healthy clip.
The problem is that the United States is using old manufacturing approaches to manage service production, and they are not working. Gains in manufacturing productivity came largely through inexorable improvements in hardware. But gains in service industry performance come largely from improvements in “humanware”-a term borrowed from the Japanese auto industry that refers to the organization of work and the skills of service managers and workers. One might expect that service firms would be leaders in humanware, but that seems not to be the case. Service firms have been even slower than manufacturing firms to adopt practices associated with high-performance work organization.
The services need a new model for improving productivity, which includes new policies to support it. For the model to take root and spread, the United States must emulate the system-building that helped generate prosperity in the past. The new economy demands a new institutional framework-a New Deal for the service economy-that meshes with service jobs and industries in the same way that the post-World War II framework suited the manufacturing economy. Several federal and state public policy initiatives could jump start the nation toward productivity gains in the service economy.
These new policies could lead the way to a new “learning economy” that will sustain U.S. prosperity in the 21st century. This learning economy would systematically and continuously promote improvement in service workers’ abilities. The alternative is a future in which a minority of well educated, highly-skilled workers monopolize the gains from a slowly growing economic pie, while too many Americans cycle among jobs that pay relatively little and offer limited prospects for advancement.
Finding competitive advantage
The service sector is much larger than most people realize. It includes transportation, communication, and utilities; finance, insurance, and real estate; retailing; professional services; and public administration-in short, everything other than goods-producing industries. As in manufacturing, higher productivity provides the foundation for higher wages. Productivity growth in the service economy, however, has averaged barely more than 1 percent per year, whereas productivity growth in manufacturing has remained at about 2.5 percent annually in recent years. Not surprisingly, median wages in services ($10 per hour in 1996) lag those in manufacturing (about $11.50).
To understand how productivity might improve, we must look closely at how services are produced. Production systems have three basic elements: hardware, software, and humanware. Hardware consists of equipment, machinery, and computers of all types. Software includes applications and systems programs. Humanware refers to the social system of production -the organization of work, management, and the skills of the labor force.
Whereas most goods-producing industries depend on specialized hardware, the hardware in much of the service sector, notably computers, is generic. Because the same technology is readily available and widely used by others, it is difficult to translate that technology into competitive advantage. Banks, for example, are rarely able to differentiate their services on the basis of their automatic teller machines. Grocery stores rarely differentiate their services because of barcode scanners. In the service sector, the differences among companies are found largely in humanware.
In contrast to manufacturers such as Alcoa, which established its dominant position in the aluminum industry through proprietary technology, airlines or hospitals buy equipment on a more-or-less turnkey basis. A hospital may differentiate its services by offering unique capabilities such as heart transplants, but it is not the transplant hardware that sways a consumer, it is the physician’s expertise. Even where custom hardware is found in the services, it is rarely as necessary for production as a good blast furnace is for producing steel.
Service organizations may also use software that can be highly specialized, such as airline reservation systems, the order-entry system for lab tests in a hospital, or the routing and scheduling algorithms used by trucking companies. But the computers on which these programs run are universal machines, and with the rise of shrink-wrapped software for ever-more-specialized functions, these once custom capabilities are also becoming commoditized. Although software may still provide a service company with some competitive advantage, it is lessening in importance.
With hardware and software being less of a distinguishing factor between successful and unsuccessful service organizations, humanware is elevated in importance. Despite this, service firms have been slower than manufacturing firms in adopting practices associated with innovative work organization and human resource management.
One reason for slow adoption is that few service firms face the kind of foreign competition that has been common in manufacturing since the late 1960s. Although domestic competition has become more intense, much of the change nationally, such as the rise of managed health care or retail “category killers” such as Home Depot, has been relatively recent. Furthermore, local firms that provide face-to-face services that must be consumed and produced in the same place get some protection because of geography. Yet firms we have studied (including a building supply company, two insurers, and a major credit card issuer) are succeeding and distinguishing themselves largely because of the attention they pay to humanware; specifically, the organization of work, training, and the application of information technology. As competition expands, especially as the World Wide Web makes it easier for foreign competitors to offer services locally, improving productivity with better humanware will become even more important.
A new productivity model
In the manufacturing era, performance improvement was driven by application of an “engineering model.” This has two major elements: the definition of a product (the chemistry of a grade of steel or the design of an electronic circuit) and specifications fixed in advance of production, and the application of technology to make a finished product that conforms as closely as possible to the specifications at the least cost. Production can be viewed as the (often very complex) solution to a technical engineering problem. In this mass manufacturing, “scientific management” generated steady improvement through a highly refined division of labor coupled with specialized hardware and software.
Recent innovations in high-performance work organization, such as total quality management and self-managed teams, partially reverse this dynamic by giving workers more discretion and responsibility. Yet they remain anchored in the scientific management tradition.
Although the engineering model is applicable to some standardized production processes in service industries, the basic assumption of a well-defined product with attributes independent of the production process applies only partially, poorly, or not at all to other service processes. In most services, the “product” differs depending on the customer: a nurse’s patient, a teacher’s student, a waitress’s diner. For each provider, a slightly or largely different process-a different model of production-applies from one customer to the next. Each process is interpretive, depending on a customer’s desires or the needs of the situation-the idiosyncracies of the copier being repaired, the mysteries of the computer program that won’t function, the specifics of the legal case. In contrast to the engineering model, in which production operations are specified through blueprints or other exact scripts, there is a substantial discretionary component in the interpretive model. Product definition and production occur simultaneously and are interdependent.
Succeeding in this environment depends largely on humanware. In the interpretive model, workers first develop an initial understanding of customer needs or the needs of the situation. They then translate that understanding into the service provided (a haircut, a legal brief, an advertising campaign). As the service begins to be delivered, they modify the services or method of delivery or interpretation of the customer’s needs. Over time, performance gains follow from improvement in the ability of workers, individually or collectively, to elicit, understand, and respond to a situation; to select and follow work practices from an available repertoire; and to learn or invent new practices as required.
Medical diagnosis and treatment is the exemplary case. Through dialogue with the patient, examination, and perhaps specialized tests or consultations with specialists, the physician explores symptoms, elicits a medical history, develops a tentative understanding of the patient’s condition, and seeks to verify and if necessary correct that diagnosis. Subjective judgments by physician and patient are part of the process, as the patient collaborates by describing or recalling symptoms and his or her history. Treatment may lead to further detective work and perhaps a change in diagnosis and an altered treatment regimen (which the patient may or may not follow). The goal, sometimes achieved and sometimes not, is to bring symptoms and treatment into congruence.
Medicine illustrates the interpretive model in its most complete form, but interpretive skills are just as important in many service jobs that do not require high levels of formal schooling or training. Even basic services call for similar interactions: helping a customer select telephone services, financial planning for a couple approaching retirement, troubleshooting the local area network in an office. In these cases too, diagnosis and treatment are intertwined. Iteration and feedback, often in real time, are essential, and the product or end point may change many times. In other cases, such as a fast food restaurant or telemarketers who follow a script, production may combine features of the engineering and interpretative models.
Because service products are so individualized, performance by the service provider can be difficult to gauge in terms of productivity. Many managers in service firms still think reflexively in engineering model terms, whether or not this is appropriate to their operation. Managed health care, with its reliance on accounting measures and decisionmaking hierarchies, follows the engineering model. This is not surprising, because there are no widely accepted measures of wellness. For example, when we visited managers in different hospitals, we were surprised to find that they invariably responded to questions about performance measures by referring to surveys of patient satisfaction (how’s the food?) or to vague future plans for collecting and analyzing data from medical records.
Likewise, much of what is meant today by terms such as data mining, knowledge management, and enterprise intelligence connotes little more than formulas derived from the old engineering model: Simplify and standardize, manage and manipulate, keep the tasks simple so anyone can do them. Such approaches may help sell credit cards or telephone calling plans. But they quickly bump up against fundamental limitations when service products are nonstandard. In one insurance company we studied, each of 70,000 business customers can have disability policies tailored specifically to their needs. The company’s workers must translate these wishes into the technical language of a policy and set an appropriate price. Heterogeneous and multidimensional service products cannot be viewed in terms of the “engineering model” associated with manufacturing.
Economies of depth and coordination
The interpretive model does not solve the measurement problem for service companies, but it does indicate how service productivity can be improved. There are two complementary pathways to performance gains: economies of depth and economies of coordination.
When workers or groups of workers improve their skills in interpreting and responding to situational needs, economies of depth result. When two or more people mutually adjust their efforts in order to define and achieve a common goal (as when nurses and physicians collaborate about a patient), economies of coordination result. Economies of depth and coordination are the principal means of improving performance in much of the service sector. Hardware and software improvements play supporting roles.
Formal education contributes to economies of depth, but competence depends heavily on experience. Individuals build up their know-how and skill incrementally and iteratively through trial and error and trial and success, as they move from school to work and from one set of tasks to the next. Research in cognitive science indicates that achieving high levels of expertise in any demanding occupation or avocation, whether radiology or chess, takes something like a decade. Over this period, the learner develops a store of previously encountered problems, patterns, good and bad solutions, rules of thumb, and heuristics from which he or she can draw when encountering a new problem.
When the lessons of experience can be passed to others, depth-related benefits multiply. For example, team meetings at insurance companies, during which policy workers discuss difficult or unusual cases, help spread economies of depth. Despite claims about “artificial intelligence,” insurers can automate only routine underwriting with knowledge-based systems that embody the accumulated experience of skilled underwriters. At present, the software cannot match senior underwriters in assessing risks and determining pricing.
A steadily growing fraction of the workforce finds itself employed in an interpretive context. Even though many of these jobs are relatively low-skilled, as in much of retail sales, they cannot be effectively automated. As a manager at a large New York bank put it: “More [careers] are going to be geared toward the analytical. The technology will accommodate the operational aspects. Looking forward, you’ll be left with a human being making a decision on extending credit when the computer goes through agency criteria and still can’t make the decision. Then there’s the creativity part of it. Getting someone to use your credit card, instead of one of a hundred others . . . You’ll still need people.”
Economies of coordination will help in many production settings where workers must mesh their efforts to achieve a common goal. They may be part of a small team, as in a restaurant, or a loose aggregation of people working for different organizations, as in a distribution network. Economies of coordination result when the ability of a work group or network to function as a unit improves. Gains may come from faster, more accurate communication and decisionmaking, sharing of tasks within and among multiskilled work groups, and processes of continuous improvement that are invisible to untrained observers (as in surgical teams). Although a bit of the gain will stem from better communications hardware and software, most of it will come from improved work practices.
Policy for a learning economy
Economic growth accompanied by steady increases in wages and living standards requires continuous growth in labor productivity. The path to greater productivity in manufacturing has been well marked: Companies rationalize production by subdividing labor processes, then mechanize and automate operations where this is cost-effective. Rationalization and mechanization lower costs and thus consumer prices. With lower prices, the market expands, allowing further rationalization and automation. This cumulative process continues to generate reasonably strong productivity growth in U.S. manufacturing.
The experience of the services has been poor by contrast. Often, gains seem to be one-time or sporadic. One food distribution company we visited had recently started tracking basic indicators of wholesaler and manufacturer performance (such as on-time delivery), and had achieved a few easy and inexpensive gains. But there was little indication that management knew what to do next. No foundation for continuous improvement had been put in place.
By applying an interpretive model, instead of an engineering model, cumulative gains in service productivity are possible. What is needed are policies that will foster economies of depth and economies of coordination. Putting such policies in place is the first step toward what might be called a learning economy. Because interpretive skills are essential throughout most of the service sector, a learning economy would be one that systematically and continuously promoted improvement in workers’ interpretive abilities, regardless of occupation or level of responsibility.
Formal schooling is a part of the foundation. All Americans should have opportunities to pursue education throughout their lives. But a learning economy is not necessarily one in which most people would have two or four years of college. Because so much of the learning in the interpretative model is experiential, Americans need richer opportunities to learn in the workplace.
Service workers must also be able to advance as a result of learning and experience. In the old economy, learning and advancement went together. Large hierarchical firms such as AT&T, Caterpillar, General Motors, and IBM provided at least the implicit promise of well-developed job ladders and long-term employment. Banks and department stores also invested in their employees. Before deregulation of financial services encroached on safe havens in local markets, banks were full of vice presidents who had started as tellers or platform workers. In the days before competition from discounters and category killers, a salesperson at Macy’s might become a buyer.
Those days are gone. Few companies of any size in any industry provide training for nonprofessional, nonmanagerial workers, other than that for immediate job tasks. The disincentives are especially strong in services. Service firms and establishments are considerably smaller than in manufacturing (averaging 14 people per establishment, as compared with 47 in manufacturing). Business networks such as health care are more fluid, and annual worker turnover in services sometimes exceeds 100 percent, as it does in nursing homes. In such settings, performance improvement is likely to be slow or nonexistent without institutions outside the firm that can support careers as well as make workers more productive over time.
Because society as a whole, not just employers, benefits from performance improvement, it makes sense for society to support the propagation of economies of depth and coordination. If workers can communicate their knowledge across organizational boundaries, benefits will spread more widely. Individuals, even those working in seemingly identical jobs, will accumulate differing stocks of knowledge; some physicians will have seen hundreds of cases of appendicitis, others dozens. Thus they need to share their experience. Some professional workers develop their economies of depth in large part through their associations- physicians and lawyers share information, mentor younger colleagues, steer friends and acquaintances to jobs. For other professionals, however, including teachers at primary and secondary levels, such mechanisms are poorly developed and knowledge diffusion is slow. For many other service workers, occupational communities are almost nonexistent. To fully exploit economies of depth and coordination, workers must be able to exchange the lessons of success and failure. Because institutions for promoting economies of depth and helping workers build fulfilling careers are underdeveloped in so many service industries, the potential payoffs are high.
Diffusion of know-how across company boundaries is especially important in an economy of smaller firms. Few small companies can give their workers the training and support needed to achieve economies of depth, if only because they lack a critical mass of employees for delivering training effectively. Small companies are also likely to be wary about putting workers in direct contact with counterparts at other companies for training, because they are afraid their employees may unwittingly give away know-how that could help a competitor. Still, service employers have less to lose than manufacturers, because humanware is not as subject to reverse engineering. One building supply company we studied, the Wolf Organization, has successfully combined training, information technology, work organization, and profit-sharing. It realizes that another company can learn what the Wolf Organization does well without knowing where to start in copying Wolf (see sidebar Leveraging Information Technology). Furthermore, like any good learning organization, Wolf has figured out how to be a better borrower than its rivals.
Avenues of learning that can strengthen interpretive skills range from apprenticeships to occupational conferences (in cyberspace as well as face to face). Industry associations could become an important vehicle. In the United States, they have often been perceived primarily as interest groups seeking to influence government through lobbying, but many business groups already play a substantial role in helping their members improve performance.
One example of how associations could do more comes from the food industry. Over the past half decade, retail grocery chains, their suppliers, and leading food manufacturers have launched a movement called “efficient customer response”: their version of manufacturing’s “just-in-time” and “quick response” practices. Their goal is to keep pace with food warehouse stores and other food discounters. A half dozen industry associations cooperated in the development of methods to increase labor productivity in trucking, warehousing, and distribution; to speed restocking of stores; and to increase inventory turns. Teams working on pilot projects drew members from manufacturers, distributors, and retailers.
Other associations have also supported innovation. In Pennsylvania, about 40 county nursing homes that are members of a statewide association hold quarterly meetings at which administrators and directors of nursing can compare notes, enhancing prospects that they will collectively challenge prevailing assumptions that nursing home work cannot be performed in any new way and hence cannot improve. The Wolf Organization is part of a nationwide group of 16 building supply companies that meet for several days at regular intervals, in part to benchmark against one another.
Each of these cases is unusual. The cooperating food stores and distributors were afraid of discounters. The nursing homes are publicly owned and face pressure to provide quality care to the low-income elderly in their communities. The regional markets of the building supply companies do not overlap greatly, so competition does not interfere with cooperation. Where firms may be reluctant to share knowledge, more of the burden for improving performance will fall on professional societies and occupational associations. They can develop consensus on best practices and help members improve their own abilities through mentoring as well as formal credential programs.
Modernizing public policy
The few cooperative efforts under way to improve economies of depth and coordination in service industries, and the many more that could develop, would be greatly accelerated by modernizing public policy. At a minimum, public policy should encourage industry and occupational associations. The U.S. Department of Commerce’s manufacturing extension centers could be adapted to the needs of service firms. Occupational associations, multifirm training, and best-practice partnerships of employers and worker representatives could be supported with seed money from government employment and training budgets. Support for national R&D to improve service industries would help as well (see sidebar National R&D Needed in Services).
Another step would be to shift the emphasis of the Commerce Department’s Malcolm Baldrige awards. These awards have helped influence what leading-edge companies regard as best practices, but although service firms are eligible, the awards go mostly to manufacturers. Indeed, the award criteria have been shaped by the engineering model and pay little attention to the interpretive model. And because only companies are eligible, the awards cannot recognize multiemployer institutions for their contributions to performance.
Federal and state governments can also help by making small grants to document the ways in which exemplary multiemployer institutions help improve performance among their members. After accumulating examples of excellence in this endeavor, government should diffuse the results and develop awards for subsequent successful applications.
Government at all levels can do still more to encourage productivity gains in the service economy. Fundamentally, what must change is the country’s outlook on where to apply assistance. The business press speaks frequently of learning organizations, but this label does not capture the possibilities inherent in the new economy. The label combines an appreciation of the importance of workers’ knowledge with a presumption, rooted in the old economy, that performance and hence prosperity depends on what happens inside the individual firm (implicitly, inside big firms). But in a dynamic service economy, performance and prosperity depend just as much on the institutions that link companies to companies in similar businesses and workers to workers who have similar jobs and expertise.
When policies are put in place to achieve this level of interaction, we will move from a set of learning organizations to an actual learning economy. That will be the New Deal for services. U.S. workers in the service industry will enjoy better-paying jobs and career advancement. They will steadily raise productivity at rates that will propel this country forward. The economic health of the United States will lead the world into the 21st century.
Eileen Appelbaum and Rosemary Batt, The New American Workplace: Transforming Work Systems in the United States. Ithaca, NY: ILR Press/Cornell University Press, 1994.
Stephen R. Barley and Julian E. Orr, eds., Between Craft and Science. Ithaca, NY: ILR Press/Cornell University Press, 1997.
Stephen A. Herzenberg, John A. Alic, and Howard Wial, New Rules for a New Economy: Employment and Opportunity in Postindustrial America. Ithaca, NY: ILR Press/Cornell University Press, 1998.
Keystone Research Center, “Technology and Industrial Performance in the Service Sector: Final Comparative Assessment,” report prepared for the U.S. Department of Commerce, Technology Administration. Harrisburg, PA: Keystone Research Center, 1998.
Stephen Herzenberg is the executive director of the Keystone Research Center in Harrisburg, Pennsylvania. John Alic is an adjunct faculty member at the Johns Hopkins School of Advanced International Studies; from 1979 until 1995, he directed numerous studies at the congressional Office of Technology Assessment. Howard Wial is a senior fellow at the Keystone Research Center and an adjunct faculty member at Rutgers University-Camden.