An Empirical Study of Flexibility in Manufacturing

Much has been written in recent years about flexible factories and flexible manufacturing systems (FMS), but the literature has been largely theoretical; managers who are interested in making their factories more flexible have little empirical research on which to base their decisions. In particular, a number of questions have yet to be answered: What are the types of flexibility that affect a company’s competitive position? How can different types of flexibility be achieved? What kinds of tradeoffs must managers make between flexibility and productivity, quality, or other performance dimensions?

To address some of these questions, we studied thirty-one printed circuit board (PCB) plants belonging to fourteen electronics firms in the United States, Japan, and Europe.1 Although our data originates from only one industry, we believe they have important implications for manufacturers elsewhere. Specifically, our research has implications for plant automation, worker participation, relationships with suppliers, wage schemes, and component reusability. We found significant relationships among different types of flexibility and discovered that increased flexibility in certain areas had no adverse quality and cost effects. In this paper, we propose a framework for incorporating flexibility into mainstream strategy analysis, describe our research, and explain our findings.

Flexibility and Strategy

The literature on manufacturing flexibility that we used as background is divided into two areas: analytical models and empirical studies. The analytical models have come almost exclusively from the fields of operations research and operations management. According to Fine’s classification scheme, there have been four main concerns in the modeling literature: (1) flexibility and life cycle theory, (2) flexibility as a hedge against uncertainty, (3) interactions between flexibility and inventory, and (4) flexibility as a strategic variable that influences competitors’ actions.2

We divide the empirical literature into four groups. The first is concerned with developing taxonomies of flexibility and is represented by the work of Gerwin; Buzacott; Mandelbaum; Browne; Slack; Kumar and Kumar; and Zelanovic.3 The second group deals with the relationship between flexibility and performance and includes work by Jaikumar; Tombak; Tombak and de Meyer; and Fiegenbaum and Karnani.4 The third group covers historical and economic analyses of flexibility and tends to view flexibility as an important attribute for the competitiveness of a firm, industry, or country. This group includes research by Piore and Sabel; Harrigan; Storper and Christopherson; Adler; Womack, Jones, and Roos; and Cusumano.5 The fourth group of literature reviews and strategic frameworks includes work by Sethi and Sethi; Suarez, Cusumano, and Fine; Hyun and Ahn; and Gerwin.6

Three main “strategic imperatives” have emerged during this century. With the advent of scientific management, early in the 1900s, efficiency became the key strategic imperative for many companies. Given the rapid, remarkable success of Ford Motor Co. with its focus on efficiency postulated by the scientific management school, the diffusion of this imperative was fast and widespread; soon after Ford’s successful growth and transformation, many companies in many sectors began to focus on improving efficiency in their operations.

Around the middle of the century, particularly after W. Edwards Deming and Joseph Juran visited Japan, quality emerged as a new strategic imperative in the marketplace. The market not only valued efficiency and low prices, but started to care more directly about the quality of products and services. As we know from the history of the quality movement, managers took longer to understand the importance of quality than the importance of efficiency (the quality movement can be traced to the early 1900s in the United States). Deming’s and Juran’s visits to Japan, and the willingness of Japanese managers to hear and apply their suggestions, started a slow process by which quality gradually became the second strategic imperative of the 1900s; only in the past two decades has the concept gained widespread attention.

The third strategic imperative, flexibility, emerged as a result of the instabilities of the 1970s and the increased global competition in most world markets starting in the early 1980s.7 Increased competition means more volatile markets, shorter life cycles, and more sophisticated buyers, which have all contributed to flexibility’s emergence as a new strategic imperative. Later in the paper, we discuss flexibility and its relevant types in more detail.

One can consider a company’s competitiveness by its position on each of the three strategic imperatives vis-à-vis its market requirements (see Figure 1). All that a company does along different dimensions (at the left side of the figure) translates into specific levels of efficiency, quality, and flexibility at each point in time. On the other side, the market (at the right of the figure), which includes not only the consumers and buyers but also competitors and regulations, demands specific levels of the three imperatives. For instance, the new European ISO 9000 requirements place tougher quality demands on firms than before; the emergence of high-quality competitors in the market would have the same effect. It is the fit between what the company achieves and what the market demands that determines a company’s competitiveness. Our framework allows for situations where all three strategic imperatives are important for competitiveness; as this study shows, trade-offs do not necessarily exist among efficiency, quality, and flexibility.

The PCB Industry

The PCB industry is divided into two major groups of players: captive plants (which we studied) that produce for downstream plants or divisions of the same company, and independent plants or contract manufacturers that sell their assembly services to different companies.

The process of assembling PCBs consists primarily of placing different components on a wired “raw” board by machine or by hand. Components vary greatly in shape, technological sophistication, and process requirements; they range from simple resistors to complex microprocessors. Most plants use two basic technologies to place components on a board — through-hole placement and surface-mount technology. The latter is a newer technology that increasingly is becoming the industry standard.

Several characteristics of our research sites are worth noting. First, most of the plants are highly automated, which means that human-related factors are less relevant than in more labor-intensive industries. Second, the production process is quite standardized, so there is less variation in production processes than in other assembly industries. Third, PCB assembly is an intermediate industry that supplies assembled boards to be used in final applications; customers are manufacturing firms rather than end users. Finally, our study covered only captive plants, which tend to be more insulated from market pressures than plants competing in the open market. Captive plants may be slower in implementing different flexibility types than plants in the open market. We found it difficult to control for corporatewide resources and policies that might influence such plant behavior.

However, we decided to research the PCB industry for several reasons. PCB manufacturers were already studying flexibility issues and had collected relevant data. Managers were willing to share data on a confidential basis. And, since PCBs are used in various applications, our sample draws from numerous companies with diverse plants.

Types of Manufacturing Flexibility

Our review of the literature on manufacturing flexibility points to many different kinds of flexibility whose definitions often overlap. Because our framework is guided by strategy concerns, we suggest that there are four basic types of flexibility to consider: mix, new product, volume, and delivery time flexibility. We consider all other types of flexibility as variants of these four basic types. For reasons of data availability, we concentrate here on the first three, which we define as follows:

• Mix Flexibility. This type of flexibility has typically been measured by the number of products that a system produces at any point in time. A highly flexible mix has therefore been equated with a broad product line (which has been associated with larger market share and profitability8). But we believe this measure is too simple. A plant that produces two very different products (such as a personal computer and a laptop) should have greater mix flexibility than a plant that produces two similar products (such as two personal computers that differ only in speed and RAM characteristics). We used the following four variables to measure the mix flexibility of the PCB producers: (1) the number of different board models assembled by each plant; (2) the number of different board sizes used during assembly by each plant; (3) the range of board density handled by each plant (in components per square centimeter); and (4) the number of product categories (e.g., VCRs, televisions, and stereos) in which the boards were used.
• New Product Flexibility. Rapid product introduction can give companies a real competitive advantage.9 We followed most of the studies in this area and measured new product flexibility by determining the number of months from the earliest stage of design to the first production run of a batch of salable products. Fewer months mean greater new product flexibility.
• Volume Flexibility. We define this type of flexibility as the ability to vary production with no detrimental effect on efficiency and quality. That is, volume flexibility is not the same as volume fluctuation, which may be associated with higher costs and lower quality levels.10 Instead of simply measuring fluctuations in volume over a certain period, we used a formula that took into account volume fluctuation, cost per placement, and the fraction of PCBs that required repair.

Factors That Affect Implementation

The literature we referred to earlier suggests that the following factors affect flexibility implementation: production technology, production management techniques, relationships with subcontractors and suppliers, human resource management, product development processes, and accounting and information systems. We tested the effects of all but accounting and information systems on the three types of flexibility we described above. (For a summary of our hypotheses and results regarding flexibility implementation, see Table 1.)

· Production Technology.

We expected companies using automated programmable technologies to have greater mix and new product flexibility. Modern PCB assembly equipment is highly programmable and therefore can be used, theoretically, to make a variety of board designs without much cost penalty. New equipment can also be linked to other parts of the factory, such as design and procurement, which should make it easier for a plant to handle a greater mix and bring out new products more rapidly. Prototyping, for example, should be easier if one can quickly program machines for new jobs.

In fact, we found that newer, more automated processes tended to be associated with less mix and new product flexibility. This is because the companies in our sample were using programmable equipment to run the largest production batches rather than to increase flexibility.11 This fact may also explain why more automated production technology correlated with greater volume flexibility.

· Production Management Techniques.

We expected “Japanese” or “lean” production management techniques to affect mix and new product flexibility. Studies have hown that these techniques tend to reduce machine setup times and improve worker involvement in the production process, characteristics that should make it easier for a company to handle a more complex product mix and to introduce new products faster.12

These management techniques did correlate with mix and new product flexibility. For instance, firms in which a high percentage of workers were involved in formal problem-solving group activities, such as quality circles, had greater mix and new product flexibility. It makes sense that such activities would broaden workers’ knowledge and allow them to understand and adapt to new tasks better.

· Relationships with Suppliers and Subcontractors.

The study confirmed our assumption that a close relationship with suppliers and subcontractors positively affects all three types of flexibility. A plant that can subcontract orders or models for which it has no adequate in-house capability can increase product variety, speed prototyping, or increase volume over in-house capacity without much cost penalty. Closeness to suppliers helps a plant to procure the right components when needed for assembly or prototyping. Otherwise, procurement may become problematic as product variety increases or new products are introduced frequently.

· Human Resource Management.

We expected human resource management to be strongly related only to volume flexibility. Even though the PCB assembly industry is relatively automated, making the human factor less important, the theoretical link between this factor and volume flexibility is so clear that we expected it to have a significant effect anyway. Plants that tend to use temporary workers should have an advantage in adjusting the work force to volume changes. Plants with wage structures linked to plant or division performance should also have an advantage in adapting to changing volume: in periods of low sales volume, the payroll burden will automatically be lowered.

We found support for the latter hypothesis. Plants with flexible wage schemes did have a cost advantage, in terms of volume flexibility, over plants with fixed wages.

· Product Development Process.

Plants that followed design for manufacturability (DFM) principles (i.e., had policies to reuse components across board models) had greater mix and new product flexibility. Higher component reusability seems to allow plants to handle a greater variety of both existing and new products. This finding may have important policy implications for plants that need to deal with increasing product variety or rapid product development, because component reusability can be promoted by specific policies, such as lists of preferred components and other incentive schemes.

To summarize, we found the following relationships between implementation factors and flexibility types:

• More automated plants tend to be less flexible, despite the programmable nature of most equipment users in the PCB industry.
• Nontechnology factors — high worker involvement, close relationships with suppliers, and flexible wage schemes —appear to increase mix, volume, and new product flexibility.
• Component reusability appears to raise both mix and new product flexibility.

Making Trade-offs with Cost and Quality

It is not always beneficial to increase flexibility, of course, if it also increases costs or decreases quality, the other two strategic imperatives in our framework. We analyzed the relationships among cost, quality, and mix and new product flexibility. (Note that our measure of volume flexibility already takes cost and quality into account.)

We measured quality in our survey sample in two ways: as the number of nonrepairable boards per million at the post-assembly check and as the percentage of boards that underwent some repair through the assembly process. Both measures were necessary because some plants have low post-assembly defect figures but do extensive board repair during assembly. Quality figures varied substantially. For example, defects per million at the post-assembly check varied from zero to 14,000.

We found no relationship between quality and flexibility. That is, plants with greater mix or new product flexibility did not have greater numbers of defects than plants with less flexibility.

Likewise, we found no relationship between cost and flexibility. Companies were unwilling to give us detailed accounts of each plant’s cost structure, so we had to rely on a common industry measure: cost per component placed. These figures varied widely, from less than $.01 to roughly $.40. Nonetheless, none of the correlations between cost and flexibility were significant.

Thus we do not detect any significant overall tradeoffs between mix or new product flexibility and either cost or quality. This seems to be in line with studies of automobiles and air conditioners that suggest that high quality, instead of being costly, is often associated with low costs or high levels of productivity, and that improvements in mix flexibility do not seem to increase costs or worsen quality levels to any great extent.13

Relationships between Flexibility Types

Different flexibility types tend to be achieved through different configurations of and emphases on production technology, production management techniques, relationships with suppliers, human resource management, and product development processes. If the different flexibility types require different configurations on all these factors, perhaps it is very difficult to achieve all of them at once. A truly flexible plant — that is, a plant that is flexible on all dimensions — may be impossible to achieve, even though the term “flexible factory” is widely used.

But, while one of our objectives was to see if there are trade-offs between two or more types of flexibility, during the data analysis, we began to see relatedness instead of trade-offs. Flexibility types seem to relate to each other rather than to work against each other. These interrelationships have various implications for manufacturing strategy.

· Mix Flexibility and Volume Fluctuation.

The literature to date has not been very careful in distinguishing between volume fluctuation and volume flexibility. Therefore, it’s worth considering volume fluctuation’s significant relationship to mix flexibility.

The plants in our study with the most mix flexibility had the lowest volume fluctuations. Thus plants that are able to achieve greater mix flexibility may enjoy the benefits of a more stable production flow. This is mainly the result of the “cushion” effect provided by the broader mix — the old story of not keeping all your eggs in one basket. For instance, plants that can switch among PCBs for many product variations will not be so adversely affected if the demand for one product line shrinks unexpectedly.

By extrapolation, all the factors that increase mix flexibility will tend to increase production stability. Closeness to suppliers and subcontractors, for example, tends to have a stabilizing effect on production volume. Not only will such relationships increase mix flexibility and therefore give the plant a cushion, but the subcontractors themselves may be a source of cushion, as firms increase or decrease subcontracting in response to demand fluctuations.

· Mix Flexibility and Volume Flexibility.

When we analyzed the relationship between mix and volume flexibility, we found none. This is not surprising, given that the factors associated with mix flexibility (e.g., lean production management techniques) are different from those associated with volume flexibility (e.g., human resource policies). However, we can think of a possible association between the two types. Plants with less mix flexibility have greater volume fluctuation and therefore need more volume flexibility. Conversely, plants with more mix flexibility may not need volume flexibility. We did not find a negative correlation such as this, but the theoretical possibility deserves further study.

· Volume Flexibility and New Product Flexibility.

Our results show a weak or nonsignificant correlation between these two types of flexibility.

· Mix Flexibility and New Product Flexibility.

These types of flexibility appear to go hand in hand. Several plants in the sample had great flexibility in both mix and new product introduction; other plants seemed to become increasingly less flexible on both dimensions. In fact, those two flexibility types tend to reinforce each other. The same factors that correlate highly with one also correlate highly with the other. For example, component reusability increases mix flexibility and shortens design cycle time. Similarly, worker involvement in problem-solving group activities is important for both types.

Consider the implications of this mutually reinforcing relationship. Some authors have championed the idea of a “focused factory,” one that has trimmed down its product variety to specialize in a narrower product line.14 Our results show that this policy may have consequences not only for mix flexibility (which is reduced almost by definition when a plant gets more focused), but also for new product flexibility. Plants that have decided to produce only a few products may be implicitly sacrificing new product flexibility and leaving themselves open to volume fluctuations. In the long term, this may jeopardize a plant’s ability to maintain high levels of capacity utilization and thus to operate profitably.

Conversely, plants that stress rapid new product introduction will naturally tend to increase their mix flexibility as time goes by, assuming the products’ rate of obsolescence is not too high. This dynamic, in turn, will tend to smooth production volume fluctuations, due to the cushion effect of mix flexibility on volume fluctuations. Thus the relationships among these three flexibility types may have powerful consequences for long-term plant performance.

Implications for Managers

In this study, we have attempted to provide both a framework and empirical evidence to help managers understand the dynamics of competitiveness and improve the position of their organizations by putting a special emphasis on managing flexibility. It is hard to manage flexibility without a framework or empirical data to guide the analysis.

By proposing a taxonomy of flexibility and collecting data on the different types of flexibility, we have shown that, despite the popularity of FMS in the literature, flexibility in our sample has much more to do with non-technology factors than with technology itself. Factors such as worker involvement in problem-solving groups, supplier relationships, and extent of component reusability in the product design process were far more important in determining the flexibility of plants in the study. In fact, the plants with more programmable automation ended up being the less flexible plants. This finding has important implications for the way an organization uses resources, and the type of investment necessary to improve flexibility. Before buying the latest available technology, managers would be better off concentrating on maximizing the potential flexibility in their current organization and technology. At least in the short term, it seems possible for managers to squeeze a lot of flexibility from their existing equipment.

Our study also shows that an organization can be very flexible in some ways and less flexible in others; thus it is not entirely appropriate to talk simply of a “flexible system.” Moreover, our results suggest that some types of flexibility tend to move together, such as mix and new product flexibility, whereas volume flexibility responds to a whole different dynamic. This again has interesting implications; managers probably have to choose the dimensions along which they want their plant to be flexible. Depending on the environment it faces, an organization may require more of certain types of flexibility than others. This is an important strategic choice that is seldom considered.

Finally, the framework we propose helps managers think of flexibility within the context of the other parameters affecting competitiveness, i.e., efficiency and quality. This allows us to arrive at a more comprehensive strategic framework that appears better suited to business today. Our three strategic imperatives framework is useful for two reasons. First, it expands the generic strategies framework by including a parameter that is very important in today’s markets and by eliminating the emphasis on trade-offs among the imperatives. Second, it creates a bridge between the literature on flexibility and mainstream strategy frameworks. Having flexibility included in the strategy concepts and frameworks will help managers be more sensitive to the importance and challenges of managing flexibility. There is no question that today’s world demands more flexibility; the real issue is being able to understand and manage it strategically.