The Merit of Making Things Fast
AN AUTO PARTS MANUFACTURER in Northern Europe started to slip in profitability. For historical reasons, manufacturing was organized in a very disjointed way with facilities scattered over the countryside, each managed independently and concentrating on only a part of the full production process. Consolidation seemed advisable to regain profitability and the managing director pondered which of several organizing themes he might adopt to guide the unit’s retrenchment.
- A manufacturing manager with a midwestern specialty chemicals operation wondered how he could revitalize his company’s manufacturing function. The time seemed ripe: the flow of products out of R&D was slowing to a trickle, and the level of the company’s inventories was beginning to alarm top management. While the principles of “just-in-time” (JIT) manufacturing appealed to him, the manager was uncertain how to apply them to his operation, since the layout was inflexible and it wasn’t possible to use different equipment.
- A Swiss precision machine maker was eager to involve the workforce to improve production, but did not know how to draw on the knowledge and energy of teams that were already established. The workforce had seen program after program come and go and were understandably skeptical.
The technologies and characteristics of these three operations were different, but in each case the managers arrived at the same solution—the reduction of throughput time. Also called cycle time, lead time, and manufacturing interval, throughput time refers to the calendar time it takes to make a product, from the time materials arrive at the factory and are available to be worked on until the finished product is awaiting shipment to a customer. How can the concept of reducing throughput time be useful? Let’s return to our examples.
- The European auto parts manufacturer decided to continue to make its requirements rather than buy them from outside suppliers, but to do so at a central site and with an entirely revised management system. Production planning was revamped, the layout rearranged around manufacturing cells, and total manufacturing area and inventories slashed in half. Throughput times shrank from twelve weeks to three, and profitability returned despite level sales.
- The midwestern specialty chemicals manager challenged his group to identify the common features of the various process steps they oversaw so that families of products could be grouped together. Department-by-department supervision was abandoned. Instead, supervisors were given responsibility for the flow of selected product families across departmental boundaries and were measured on how quickly they could shepherd batches of product through the entire process. The result: inventories went down, quality went up, the workforce was enthusiastic, and throughput times plunged from forty-five to five days
- The Swiss machinery company charged its teams to find ways to shorten the time materials spent in their workstations. Management was delighted with the workforce’s quick and enthusiastic response.
These are not isolated tales. Many leading firms have instituted programs that focus on throughput time reduction. Among them:
- IBM’s Continuous Flow Manufacturing;
- Westinghouse’s OPTIM (Operating Profit through Time and Investment Management), which explicitly tracks value added over time so that the impact of throughput time reduction can be seen more graphically; and
- Motorola’s Short Cycle Management program. Motorola has formally written throughput time reduction, together with quality improvement, into its corporate strategic goals.
In addition, numerous companies that have found success with just-in-time manufacturing principles—Deere, Black & Decker, Hewlett-Packard, Harley-Davidson, Digital Equipment, among others—now routinely track throughput time as an indicator of manufacturing progress.
A Good Choice for Manufacturing Policy
My recent research on productivity has clearly linked throughput time reduction and several other linchpins of just-in-time manufacturing to factory productivity gain. For instance, research statistics suggest that halving throughput time is worth an additional two or three percentage points to a plant’s rate of productivity gain. In addition, out of many potential means of improving productivity, only the JIT-related ones were statistically shown to be consistently effective.
These results derive from two research efforts, one confined to U.S. factories and the other involving factories from thirty different countries. The U.S. research involved two parallel studies. In the first, data from 265 plants in diverse industries nationwide was gathered in an eight-page mail survey. In the second, data on the performance and operations of 26 plants over the previous five years in two industries (computers and vehicles) was personally collected. The international study gathered data from 128 plants using a mail survey nearly identical to the U.S. one.1 (Table 1 contains summary statistics from both mail surveys.)
Since all three studies included information about each factory’s products and production processes, performance, and existing programs, it was possible to examine a number of management policies that have been considered productivity enhancing. For example, substantial information was collected about automation and new production equipment and technologies, motivation programs, value engineering, computer-based information and planning systems, gain-sharing plans, quality-of-work-life initiatives, and management reorganization, as well as JIT-related policies such as quality programs, inventory reduction, group technology and relay-out, and throughput time reduction. In addition, the data permitted me to control for such features as factory size, kind of industry or process, unionization, and location.
While many adherents of just-in-time manufacturing base their claims on anecdotal evidence, this research used large sample statistics and tested alternatives to JIT. For example, the plants with the highest productivity gains were not, in general, distinguished from those with lesser gains by heavy investment in high tech, Class A MRP II systems, gain-sharing plans, or conventional industrial engineering approaches that concentrate on the operator’s job. The only consistent distinctions were those relating to JIT, specifically, throughput time reduction, improved quality, lower inventories, and participative management techniques. On these, all three studies were unanimous.
In addition, younger plants did no better than older ones, and smaller plants no better than bigger ones. Similarly, there was no distinction between union and nonunion, sunbelt and frostbelt, Asian and non-Asian, and Northern European and Southern European plants. Moreover, the type of industry does not seem to affect the results: process and nonprocess industries do about the same. It appears that management, rather than geography, size, union status, age, or industry, holds the key to a factory’s productivity gain.
What Progressive Companies Are Tracking
These research results are in harmony with the current practices of many leading companies. Reports of manufacturing initiatives (as chronicled in trade journals like Target, the publication of the Association for Manufacturing Excellence) show that key indicators of manufacturing progress for these companies are inventory turns (or days of inventory on hand), metrics of quality such as reject rates or first-pass yields, the amount of space freed up for other uses, slashes in setup time, and throughput time reductions. Representing factors that appear to be significant for productivity gain, these measures attract attention both inside and outside these progressive companies.
These same reports show that progressive companies in many industries no longer track and report labor efficiency and machine utilization.2 Yet many companies still compute—and manage with—these measures. The use of labor efficiency (standard labor hours divided by actual labor hours) is justified as a measure of labor effort and effective scheduling. Machine utilization (hours run divided by total hours of machine availability) is presumed to measure effective scheduling and machine choice, and is often used to justify equipment acquisition. However, labor efficiency measures often stimulate foremen to keep workers busy, sometimes by producing unnecessary parts or products, and machine efficiencies provoke managers to keep needless queues of work-in-process inventory in front of machines and to skimp on preventive maintenance. Progressive companies have scuttled these measures.
Why Throughput Time Reduction Makes Sense
The obvious question to pose is this: Why do productivity and profitability increase when throughput times are collapsed? The fable of the tortoise and the hare can help our understanding. As we all recall, the tortoise won the race because it never stopped moving. Each step was slow, but the tortoise was diligent and single-minded. The hare, on the other hand, was all raw speed, but he stopped and started a lot and had a propensity to be diverted from the race.
As Table 2 illustrates, a lot of factories run like hares. They operate equipment that is fast and fancy—when it is set up and actually running. However, materials in those factories often lie unused, typically in a work-in-process inventory, like the hare that hasn’t gotten off its haunches. Setup times are long and equipment breakdowns frequent. The tortoise-type factory can actually make things faster, not because its equipment runs at higher speeds, but because materials keep moving, slowly but surely, through each process step. Value is added for a larger portion of time spent in the process. Making things fast, then, does not mean operating the fastest machines or having the most automation. It means designing and organizing the factory so that materials are always moving forward through well-maintained equipment that is easy to change over from one product to another.
How does a factory become less like a hare and more like a tortoise? The research suggests concentrating on throughput time. While throughput time reduction does not improve productivity by itself, it stimulates a host of complementary actions and tactics within the factory that, in turn, improve productivity.
The Case for Throughput Time Reduction
Consider some of the actions that the broad-gauge objective of throughput time reduction encourages.
Good Quality.
Throughput time is best defined as the weighted average time through the process of all the units of a representative batch or order. Thus, if throughput time is to be lessened, managers cannot afford to let an order languish in a rework station or sit idly waiting for scrapped output to be replaced. All of the order needs to be made right the first time if throughput time is to reach its lowest value. Feedback on problems must be quick and solutions timely. Problem-solving skills take on new importance and urgency, as does the desirability of involving everybody in the factory in the task. Throughput time reduction thus complements the push to improve quality.
Low Inventories.
Inventories are perhaps the chief culprit in hindering the speedy travel of materials through the process. Throughput time reduction stimulates a variety of means for lowering inventories. For instance, the workforce might make a bit of everything demanded every day, building in small volumes in just the quantities sold in the marketplace rather than building large inventories to tide the factory over until the next product is produced. Other means include:
- layouts that make storing work-in-process inventory inconvenient and that facilitate the movement of materials between workstations;
- setup time reduction that facilitates small-size batches;
- vendor relationships that emphasize good quality and reliable deliveries; and
- production controls that employ relatively few points of control in the factory, lessening artificial barriers to the flow of materials.
Process Rationalization.
Throughput time reduction can focus attention on which steps of the process truly add value and which do not, leading to the removal of unnecessary steps and the modification of others.
Attention to Bottlenecks.
Throughput time reduction requires the elimination of bottlenecks. One simply cannot dismiss the hard decisions involved with bottleneck breaking by observing that the squeeze on capacity will evaporate when the schedule is changed or when “the heat is off? These problems—involving quality, maintenance, layouts, flows, operator training, and the like—must be analyzed and solved, preferably permanently. Here again, the involvement of everyone in the factory assists this process.
Diminished Chaos and Confusion.
Fluctuations in the product mix and in volumes, expediting of shop orders and vendor deliveries, engineering and purchasing change orders, and materials shortages have long been recognized as detrimental to productivity gain. Concentrating on throughput time reduction, however, helps managers avoid the all-too-compelling sense of urgency—and the resulting chaos and confusion—these conditions present. It is better to work toward a lower average throughput time for representative orders than to rush around trying to expedite the “hot” one.
Next consider three key spillover benefits.
Overhead Elimination.
In many factories, overhead is the most rapidly growing cost item. Much of this can be attributed to the factory’s numerous transactions, many on the materials side of the business—purchasing, materials receipt, inspection, production planning, inventory control, materials handling, production control, and all the accounting these entail. If these transactions and accounting can be eliminated, the overhead personnel associated with them can be eliminated as well. I know of no better way to do this than to focus on throughput time reduction. The popular Pareto “80-20” rule may apply—that 80 percent of the factory’s headaches, especially for the overhead functions, come from 20 percent of the products, the 20 percent that, for one reason or another, lingers in the factory. With a quick flow of materials through the factory, many fewer resources can be spent planning for them, tracking them, adjusting their status, and costing them. Simple, quick flows within the process can breed simple, quick management systems to plan and control them.
Quick Response to the Market.
Quick delivery of exactly what the customer wants has always been an effective selling tool. Concentration on throughput time helps to align capacity and output rates to the quantities actually demanded in the marketplace, allowing manufacturing to better support the sales effort.
Improved Capital Appropriations.
Conventional justifications for new equipment focus on direct labor cost reductions, machine utilization increases, and (sometimes) quality improvements. Such practices do not take into consideration the change in throughput time that any new investment can make. Lower throughput times can enhance flexibility and achieve quicker customer service, while space and inventories decline and quality improves. Other things being equal, a capital expenditure that promises to reduce throughput time is preferable to one that does not, even if it means holding “excess capacity” (which is better than “excess inventory”). Indeed, the potential of any new investment or technology is fully realized only when it can speed up the flow of materials through the process.
The Power of Linking
The attractiveness of throughput time reduction as a principle to guide productivity improvement is powerfully illustrated by the use of manufacturing cell (group technology) concepts, U-shaped lines, and multimodel lines for particular classes of products. (The midwestern specialty chemicals maker mentioned earlier is one example.) Some years ago many plants’ processes could be described as hybrids—part batch flow and part line flow. Now significant numbers have linked batch operations directly to line flow operations. Doing this usually means altering the way the factory is organized and measured, but, as Figure 1 illustrates, results in greatly improved productivity, flexibility, space utilization, and quality.
Part A of this figure depicts a common hybrid operation. The first part of the process is fabrication, typically done in sizable batches on a series of machines, with similar machines located together. Next, a standard assembly line tries to match its production to customer orders. Since the fabrication portion of the factory is frequently viewed as too slow and clumsy to match its production rate to that of the assembly line, production plans and schedules for the two halves of the factory are different. Thus, “decoupling” inventory is necessary as a buffer when the requirements of the assembly line outstrip the production of the slower-reacting batch flow operation.
The goal of consistently reducing throughput time rejects the need for a decoupling inventory whose purpose is to keep labor efficiencies and machine utilizations high. Instead, it emphasizes setup time reduction and process simplification to improve the reaction time of the fabrication operation and to link it directly—both physically and through production planning—to the assembly line. The hybrid process thus gets converted to a more powerful line flow operation (part B, Figure 1) that enhances productivity and flexibility. While particular departments may suffer by conventional measures, it is better to have idle but flexible labor that produces just what the market is buying than to keep labor busy producing what is not really needed.
Something for Everyone
Throughput time reduction as a principle for productivity improvement in the factory has meaning for everyone, from the operator to the vice president of manufacturing. As we saw with the Swiss machinery company, reducing throughput time is an easily understood goal for the average shopfloor worker who observes every day the problems and policies that slow up the flow of materials through an area. Throughput time reduction can help make JIT concepts tangible to a confused or skeptical workforce and can help invigorate otherwise dull production jobs.
For the plant manager, reducing throughput time often means working on cross-departmental and cross-functional problems, which from his or her vantage point are easy to spot, if not necessarily easy to remedy. And, for the vice president of manufacturing and others in top management, reducing throughput time can mean reassessing plant-to-plant logistics as well as product and technology assignments. From the vantage point of top management, these may be the most compelling issues.
Throughput time reduction is an objective that cuts across many factory fiefdoms—equipment, layout, quality, materials handling, production planning, inventory control, and cost accounting—but without the potential for conflict accompanying many other departmental and factory performance measures.
As a final example of the power of this way of thinking, consider the situation of a midwestern manufacturer whose product currently takes sixteen weeks to make, despite a labor content of just two and a half hours per unit. The labor-intensive portion of the product’s manufacture, accounting for one and a half hours of labor, is accomplished south of the U.S. border, a procedure thoroughly consistent with conventional views on cost cutting. Unfortunately, the costs of transportation, handling, and inventory carrying get buried in overhead, and no one really knows what they are.
In this instance, trying to slash throughput time naturally leads to the question, Why must the product travel so far just to save direct labor expense? Revamping the flow in the U.S. factory might save as much, or more, true cost. The company is currently exploring what it would take to produce the entire product in the United States in only three or four days. The general manager is enthusiastic about the prospect of quicker reaction time in the marketplace and the resulting increased market share.
There are many techniques for improving factory productivity, and managers must be careful in making selections. This research and the experience of increasing numbers of companies suggest that contraction of throughput time and its companion tactics—improved quality, lower inventories, setup time reduction, revamped layouts, smaller lot sizes, increased stability in the production schedule, and more order in engineering changes—is a promising way to approach this complex problem. It seems that Aesop was right all along: we should put our money on the tortoise and not the hare. It makes you wonder how Aesop would have done as a plant manager.
Appendix
The conclusions about throughput time derive from two distinct research efforts. The first, confined to U.S. factories, took place from 1984 to 1986 and was funded primarily by the U.S. Department of Commerce with other funding by Control Data Corporation. The second, involving factories from thirty different countries, was sponsored by IMEDE, the international business school in Lausanne, Switzerland, and was conducted while I was a visiting professor there in 1986–1987.
The American study encompassed two databases. First, using an eight-page mail survey, I gathered 220 items of data from 265 responding plants in diverse industries nationwide. Most of these plants were large, and many belonged to sizable parent corporations. The second domestic database was derived from visits to twenty-six plants, fourteen computer plants and twelve vehicles plants (fabrication and assembly of auto/truck/farm vehicles or their components). During one- or two-day visits, I collected detailed information on these plants’ productivity performance during the previous five years as well as time-series information on a variety of plant characteristics. Information and opinions were collected from the plant manager, manufacturing and industrial engineering, personnel, accounting, production planning and control, purchasing, and quality control. Both industries chosen for plant visits are major national industries and are divergent in character. The identities of all the cooperating companies and managers are confidential.
Since the mail survey developed a broad cross section of plants, while plant visits resulted in detailed performance and time-series information from a smaller cross section of plants, the results are more general than if only one database or research technique had been used. The results of both approaches are consistent.
The international study conducted at IMEDE, employing a mail survey that differed only slightly from the one used in the United States, confirmed the universality of the results. The survey was mailed to manufacturers on IMEDE’s international mailing list, and 128 plants from thirty different countries responded. Most responses originated in Europe.
In addition to scores of questions about a plant’s products, processes, management policies, and performance, five distinct measures of productivity performance were solicited by the surveys and during plant visits. The firmest and most readily obtained was each plant’s recent gain in labor productivity. Other measures included general productivity, the plant’s perceived rank within its company and industry, and whether its productivity gains have been speeding up or slowing down over time. Differentiating statistically why some plants appear to do better than others constituted the heart of the research. The chief research tool, regression analysis, was used to control for the effects of numerous plausible influences on factory productivity. Conventional explanations of factory productivity were also thoroughly tested.
References
1. R.W. Schmenner, “Comparative Factory Productivity” (Washington, DC: research report for the Research and Evaluation Division, Economic Development Administration, U.S. Department of Commerce, July 1986);
R.W. Schmenner, “An International Examination of Factory Productivity” (Lausanne, Switzerland: Summary findings for IMEDE, June 1987).
2. In some industries, particularly those using continuous flow processes, machine utilization figures may still make considerable sense.