Predecir para prevenir: «si no está roto, no lo arregles»

During the Golden Age of mass production, Japanese companies set about improving mass production methods. The approach they developed has come to be known, variously, as ‘‘Toyotaism,’’ ‘‘lean production’’ (Womack et al. 1990; Womack and Jones 1994), and ‘‘Xexible mass production’’ (Boyer and Coriat 1986). Whatever the label, the nature of their accomplishment became apparent in the early 1970s. North America’s, and many of Europe’s, industrial systems were mired in problems of bad quality, slow productivity growth, low proWts, labor–management conXict, high unemployment, and inXation – in short, the crisis of mass production, discussed in the previous chapter. At

the same time, Japanese products – automobiles, ships, oYce machines, consumer electronics, cameras, industrial equipment, and so on – were gaining market share.

The Japanese manufacturers were producing higher quality goods, and doing so at lower cost. They did this by modifying mass production methods. They kept mass production’s focus on maintaining the Xow of materials, and also on the precise speciWcation of measurements and methods, in order to produce interchangeable parts reliably and cheaply. They made fundamental changes, however, to the inXexibility of manufacturing equipment; to quality control procedures; to the methods used to improve production processes; and to design processes for parts and sub-assemblies. In making these changes, they also changed the relationship between the company and its employees, and between the company and its suppliers. I discussed these changes brieXy in Chapter 2, and will now consider them in greater detail.

Bringing Xexibility to mass production was initially a method for coping with both the shortage of equipment and the small consumer market in Japan during the late 1940s and 1950s. Toyota sought to move from craft production to mass production, but it could not aVord such luxuries as separate press for each sheet-metal shaping oper-ation. A press can be used for many diVerent operations, but the die used in the press provides the shape for a particular part. To use the same press for multiple operations, they had to work out how to change the die quickly. Learning how to be quick was a gradual process. Beginning with American presses in the late 1940s, it took Toyota a day to change a die; by the late 1950s, it took three minutes. One of the things that made such quick changes possible was a change in the organization of work. The mass production practice was to have die-changing done by specialists, while the stamping operation was done by less skilled operatives. The Toyota practice was to have the press operators change the die themselves (Womack, Jones, and Roos 1990, pp. 50–2). In this way the use of multi-skilled production workers became central to Xexibility.

The gradual progress toward the three-minute change of die is emblematic not only of production Xexibility, but of the continuous improvement for which Toyota-style systems are known. Process improvement is about both lowering cost and improving quality, and it is convenient to discuss it together with quality control. The general approach is to involve production workers in both the identiWcation of problems and possibilities for improvement, and in the formulation of new procedures to address these problems and opportunities. At the simplest level, this meant suggestion boxes, and the recognition of workers who had suggested improvements that the management adopted. The suggestion box was a simple idea observed by Toyota engineers on a visit to Detroit in the late 1940s; like many such ideas, it was implemented more eVectively by Toyota than by those they had learned it from.

Another tool for problem identiWcation has been the systematic collection of data on defects as they show up during the production process, and analysis of that data using simple statistical techniques. Like suggestion boxes, the principles of statistical quality

134 THE GLOBAL ENVIRONMENT OF BUSINESS

control were imported to Japan from the US. The approach had been promoted by the military for American wartime production. The US Army carried on with this advocacy when it occupied Japan after the war. In the end, statistical quality control was far more enthusiastically received by Japanese manufacturers than it had been by American ones, perhaps because it worked best when implemented by skilled production workers with job security. American statisticians W. Edwards Deming and Joseph Juran Wrst became famous in Japan, and then decades later in the US.

Compared with other large Japanese manufacturers, Toyota was a latecomer to the use of statistical quality control (Fujimoto 1998). When it did embrace the technique, however, Toyota combined it with another system the company had pioneered: just-in-time (JIT) delivery of parts. In some respects, JIT would have seemed familiar to Henry Ford: Ford’s early assembly lines had been part of a lean system; one of that system’s great accomplish-ments had been to synchronize the production cycle times for parts, so that the Xow of production continued uninterrupted. But the Model T was a simple car, with far fewer parts than a modern one, and Ford made it with almost no options (‘‘any colour you want as long as it’s black’’). As mass produced products had become more complex and options had proliferated, the once-lean system had become Xabby. Toyota worked out how to synchronize the delivery of just the right parts for the models and options being produced.

For a good standard discussion of the JIT method, see Womack et al. (1990).

It is obvious that JIT can reduce inventory holding costs, but the logic linking JIT and quality management is less well understood. In any system, production workers have schedules to keep. Noticing a defect in the parts or materials they are using, they would like to meet their schedule by setting the defective item aside and reaching into inventory for a replacement. Once set aside, tracing the origins of the defect becomes a low priority.

The problem with that is that a defective part is usually not alone: it indicates a problem in the process for making the part. JIT makes it impossible to ignore defects.

One might object that there is as second way production workers can deal with defective parts, which is simply to use them and then try to hide the defect. This was a common problem under mass production. The system lacked good ways of identifying defects, tracing them to their source, and correcting the system that had produced them;

despite large inventories, there were often no really good parts or materials available.

One might expect that the problem would be even greater under JIT; with no inventory to fall back on, the pressure to accept parts as is can surely be tremendous. Yet, in practice, the lean production system has avoided this problem.

Advocates of lean production and total quality management argue that the reason this problem persisted under mass production is the Taylorist separation of conception and execution: production workers were told simply to produce, and were not enlisted in identifying and solving problems. Quality control under mass production was a matter for inspectors at the end of the production process. They would look for defects, and when these were found somebody would be blamed. For production workers, hiding defects from the inspectors became a game. Attempting to ‘‘inspect quality in’’

is held to be inherently inferior to the lean production approach of ‘‘building it in.’’ The quality accomplishments of Japanese manufacturers are evidence that employee involvement in quality management can turn the pressure imposed by JIT from a motive to hide problems to a motive to identify and solve them.

But once problems have been identiWed, how are they solved? Here, again, the Toyota-style systems depart from Taylorist mass production in fundamental ways. A problem may be assigned to a team of employees – sometimes called a ‘‘quality circle’’ – who are asked to study the problem and propose a solution. Participation by production workers in this kind of problem-solving is usually devoted to making incremental improvements in the production process. Yet though the composition of a problem-solving group for a large design project will be diVerent from one devoted to incremental improvement in produc-tion, what we see in both cases is a wider circle of collaboration than there would be under mass production. Collaboration up and down the supply chain with the system of ‘‘black box’’ design has been noted in Chapter 2. In addition to this, the Toyotaist approach has long been one of design-for-manufacture. In mass production systems, design and pro-duction engineering tended to be distinct stages; in the Toyota-style, systems have product designers and production engineers working together from the start.

The Toyota approach – skilled production workers solving problems – is seen by many as the antithesis of Taylorism. It is, and it isn’t. The tasks of Toyota’s multiskilled production workers are minutely speciWed, as in any mass production system. A solution to a production problem, after it has been proposed by a team of workers and accepted by their managers, serves to reWne those task instructions. From an engineering standpoint, the new instructions are better than the old ones, which is why Toyota was able to produce cars with fewer defects than its mass production rivals.

From the production worker’s standpoint, however, the instructions left even less discretion, and less slack time, than the mass production system. For this reason, some critics of the Toyota system call it ‘‘management by stress’’ (Parker and Slaughter 1998). It is perhaps no coincidence that the system was developed in a country where the attachment of workers to their jobs, and the common requirement to work long hours, have given the language a single word (karoshi) meaning ‘‘death by over-work.’’ The roots of this extreme devotion to the company will be considered below.