Unformatted text preview: OPERATIONS AND LOGISTICS
STRATEGIES. ___________________________ READING:
“RUNNING THE FACTORY”
Document published exclusively for this course / seminar.
It is not permitted reproduce it totally or partially without permission.
Document publicat per a ús exclusiu d’aquest curs/seminari.
està permès reproduir-lo totalment o parcialment sense permís. No Documento publicado para uso exclusivo de este curso/seminario. No está
permitido reproducirlo total o parcialmente sin permiso. Faculty: PhD. August CASANOVAS RUNNING THE FACTORY12 The automotive assembly plant dominates itslandscape, wherever in the world
it's located. From a distance, it is a vast windowless mass surrounded by acres of
storage areas and railway yards. The complex shape of the building and the lack of a
facade often make it hard to know just where to enter. Once inside, the scene is initially
Thousands of workers in one vast building tend to streams of vehicles moving
across the floor, while a complex network of conveyors and belts in the lofty ceiling
carries parts to and fro. The scene is dense, hectic, noisy. On first exposure, it’s like
finding oneself inside a Swiss watch—fascinating but incomprehensible and a little
frightening, as well.
In 1986, at the outset of the IMVP, we set out to contrast lean production with
mass production by carefully surveying as many of the world’s motor-vehicle assembly
plants as possible. In the end, we visited and systematically gathered information on
more than ninety plants in seventeen countries, or about half of the assembly capacity
of the entire world. Ours would prove one of the most comprehensive international
surveys ever undertaken in the automobile or any other industry.
Why did we choose the assembly plant for study? Why not the engine plant,
say, or the brake plant or the alternator factory? And why so many plants in so many
countries? Surely, the best lean-production plant in Japan and the worst massproduction plant in North America or Europe would have sufficiently demonstrated the
differences between lean and mass production.
Three factors convinced us that the assembly plant was the most useful activity
in the motor-vehicle production system to study.
First, a large part of the work in the auto industry involves assembly. This is so
simply because of the large number of parts in a car. Much of this assembly occurs in
components plants. For example, an alternator plant will gather from suppliers or
fabricate the 100 or so discrete parts in an alternator, then assemble them into a
complete unit. However, it's hard to understand assembly in such a plant, because the
final activity usually makes up only a small part of the total. In the final assembly plant,
by contrast, the sole activity is assembly—welding and screwing several thousand
simple parts and complex components into a finished vehicle.
1 2 The Machine that Change the World, from James P. Womack. Chapter 4: Running the Factory. This chapter is based on the IMVP World Assembly Plant Survey. The survey was initiated by John
Rrafcik, who was later joined by John Paul MacDuffie. HaruoShimadaalsoassisted. Second, assembly plants all over the world do almost exactly the same things,
because practically all of today’s cars and light trucks are built with very similar
fabrication techniques. In almost every assembly plant, about 300 stamped steel
panels are spot-welded into a complete body. Then the body is dipped and sprayed to
protect it from corrosion. Next, it is painted. Finally, thousands of mechanical parts,
electrical items, and bits of upholstery are installed inside the painted body to produce
the complete automobile. Because these tasks are so uniform, we can meaningfully
compare a plant in Japan with one in Canada, another in West Germany, and still
another in China, even though they are making cars that look very different as they
emerge from the factory.
Finally, we chose the assembly plant for study, because Japanese efforts to
spread lean production by building plants in North America and Europe initially involved
assembly plants. When we began our survey in 1986, three Japanese-managed
assembly plants were already in operation in the United States and one was ready to
open in England.
By contrast, Japanese plants for engines, brakes, alternators, and other
components, though publicly announced for North America and Europe, were still in the
planning stage. We knew from experience that it's pointless to examine a company's
blueprints for a new plant or to look at a plant just as it startsproduction. To see the full
difference between lean and mass production at the plant level, we had to compare
plants operating at full volume.
What about the second question we are often asked: "Why study so many
plants in so many countries?" The answer is simple. Lean production is now spreading
from Japan to practically every nation. Directly in its path are the giant mass-production
plants of the previous industrial era.
In every country and every company—including, we might add, in the less
accomplished companies in Japan—we have found an intense, even desperate, desire
to know the answer to two sim-ple questions: "Where do we stand?" and "What must
we do to match the new competitive level required by lean production?" Now we know
the answers. CLASSIC MASS PRODUCTION: GM FRAMINGHAM
We began our survey in 1986 at General Motors' Framingham, Massachusetts,
assembly plant, just a few miles south of our home base in Boston. We chose
Framingham, not because it was nearby, but because we strongly suspected it
embodied all the elements of classic mass production.
Our first interview with the plant's senior managers was not promising. They had
just returned from a tour of the Toyota-GM joint-venture plant (NUMMI) where John
Krafcik, our assembly-plant survey leader, formerly worked. One reported that secret
repair areas and secret inventories had to exist behind the NUMMI plant, because he hadn't seen enough of either for a "real" plant. Another manager wondered what all the
fuss was about. "They build cars just like we do." A third warned that "all that NUMMI
talk [about lean production] is not welcome around here."
Despite this cold beginning, we found the plant management enormously
helpful. All over the world, as we have since discovered again and again, managers
and workers badly want to learn where they stand and how to improve. Their fear of
just how bad things might be is in fact what often creates initial hostility.
On the plant floor, we found about what we had expected: a classic massproduction environment with its many dysfunctions. We began by looking down the
aisles next to the assembly line.They were crammed with what we term indirect
workers—workers on their way to relieve a fellow employee, machine repairers en
route to troubleshoot a problem, housekeepers, inventory runners. None of these
workers actually add value, and companies can find other ways to get their jobs done.
Next, we looked at the line itself. Next to each work station were piles—in some
cases weeks' worth—of inventory. Littered about were discarded boxes and other
temporary wrapping material. On the line itself the work was unevenly distributed, with
some workers running madly to keep up and others finding time to smoke or even read
a newspaper. In addition, at a number of points the workers seemed to be struggling to
attach poorly fitting parts to the Oldsmobile Ciera models they were building. The parts
that wouldn't fit at all were unceremoniously chucked in trash cans.
At the end of the line we found what is perhaps the best evidence of oldfashioned mass production: an enormous work area full of finished cars riddled with
defects. All these cars needed further repair before shipment, a task that can prove
enormously time-consuming and often fails to fix fully the problems now buried under
layers of parts and upholstery.
On our way back through the plant to discuss our findings with the senior
managers, we found two final signs of mass production: large buffers of finished bodies
awaiting their trip through the paint booth and from the paint booth to the final assembly
line, and massive stores of parts, many still in the railway cars in which they had been
shipped from General Motors' components plants in the Detroit area.
Finally, a word on the workforce.Dispirited is the only label that would fit.
Framingham workers had been laid off a half-dozen times since the beginning of the
American industry's crisis in 1979, and they seemed to have little hope that the plant
could long hold out against the lean-production facilities locating in the American
Midwest. CLASSIC LEAN PRODUCTION: TOYOTA TAKAOKA
Our next stop was the Toyota assembly plant at Takaoka in Toyota City. Like
Framingham (built in 1948), this is amiddle-aged facility (from 1966). It had a much
larger number of welding and painting robots in 1986 but was hardly a high-tech facility of the sort General Motors was then building for its new GM-10 models, in which
computer-guided carriers replaced the final assembly line.
The differences between Takaoka and Framingham are striking to anyone who
understands the logic of lean production. For a start, hardly anyone was in the aisles.
The armies of indirect workers so visible at GM were missing, and practically every
worker in sight was actually adding value to the car. This fact was even more apparent
because Takaoka s aisles are so narrow.
Toyota's philosophy about the amount of plant space needed for a given
production volume is just the opposite of GM s at Framingham: Toyota believes in
having as little space as possible so that face-to-face communication among workers is
easier, and there is no room to store inventories. GM, by contrast, has believed that
extra space is necessary to work on vehicles needing repairs and to store the large
inventories needed to ensure smooth production.
The final assembly line revealed further differences. Less than an hour's worth
of inventory was next to each worker at Takaoka. The parts went on more smoothly
and the work tasks were better balanced, so that every worker worked at about the
same pace. When a worker found a defective part, he—there are no women working in
Toyota plants in Japan—carefully tagged it and sent it to the quality-control area in
order to obtain a replacement part. Once in quality control, employees subjected the
part to what Toyota calls "the five why's" in which, as we explained in Chapter 2, the
reason for the defect is traced back to its ultimate cause so that it will not recur.
As we noted, each worker along the line can pull a cord just above the work
station to stop the line if any problem is found; at GM only senior managers can stop
the line for any reason other than safety—but it stops frequently due to problems with
machinery or materials delivery. At Takaoka, every worker can stop the line but the line
is almost never stopped, because problems are solved in advance and the same
problem never occurs twice. Clearly, paying relentless attention to preventing defects
has removed most of the reasons for the line to stop.
At the end of the line, the difference between lean and mass production was
even more striking. At Takaoka, we observed almost no rework area at all. Almost
every car was driven directly from the line to the boat or the trucks taking cars to the
On the way back through the plant, we observed yet other differences between
this plant and Framingham. There were practically no buffers between the welding
shop and paint booth and between paint and final assembly. And there were no parts
warehouses at all. Instead parts were delivered directly to the line at hourly intervals
from the supplier plants where they had just been made. (Indeed, our initial plant
survey form asked how many days of inventory were in the plant. A Toyota manager
politely asked whether there was an error in translation. Surely we meant minutesof
A final and striking difference with Framingham was the morale of the
workforce. The work pace was clearly harder at Takaoka, and yet there was a sense of
purposefulness, not simply of workers going through the motions with their minds elsewhere under the watchful eye of the foreman. No doubt this was in considerable
part due to the fact that all of the Takaoka workers were lifetime employees of Toyota,
with fully secure jobs in return for a full commitment to their work.1 A BOX SCORE: MASS PRODUCTION VERSUS LEAN
When the team had surveyed both plants, we began to construct a simple box
score to tell us how productive and accurate each plant was ("accurate" here means
the number of assembly defects in cars as subsequently reported by buyers).2 It was
easy to calculate a gross productivity comparison, dividing the number of hours worked
by all plant employees by the number of vehicles produced, as shown in the top line of
Figure 4.13 However, we had to make sure that each plant was performing exactly the
same tasks. Otherwise, we wouldn't be comparing apples with apples. So we devised a list of standard activities for both plants— welding of all body
panels, application of three coats of paint, installation of all parts, final inspection, and
rework—and noted any task one plant was doing that the other wasn't. For example,
Framingham did only half its own welding and obtained many prewelded assemblies
from outside contractors. We made an adjustment to reflect this fact.
We also knew it would make little sense to compare plants assembling vehicles
of grossly different sizes and with differing amounts ofoptional equipment, so we adjusted the amount of effort in each plant as if a standard vehicle of a specified size
and option content were being assembled.4
When our task was completed, an extraordinary finding emerged, as shown in
Figure 4.1. Takaoka was almost twice as productive and three times as accurate as
Framingham in performing the same set of standard activities on our standard car. In
terms of manufacturing space, it was 40 percent more efficient, and its inventories were
a tiny fraction of those at Framingham.
If you remember Figure 2.1 from Chapter 2, you might wonder whether this leap
in performance from classic mass production, as practiced by GM, to classic lean
production, as performed by Toyota, really deserves the term revolution. After all, Ford
managed to reduce direct assembly effort by a factor of nine at Highland Park.
In fact, Takaoka is in some ways an even more impressive achievement than
Fords at Highland Park, because it represents anadvance on so many dimensions. Not
only is effort cut in half and defects reduced by a factor of three, Takaoka also slashes
inventories and manufacturing space. (That is, it is both capital and labor-saving
compared with Framingham-style mass production.) What's more, Takaoka is able to
change over in a few days from one type of vehicle to the next generation of product,
while Highland Park, with its vast array of dedicated tools, was closed for months in
1927 when Ford switched from the Model T to the new Model A. Mass-production
plants continue to close for months while switching to new products. DIFFUSING LEAN PRODUCTION
Revolutions in manufacture are useful only if they are available to everyone. We
were therefore vitally interested in learning if the new transplant facilities being opened
in North America and Europe could actually institute lean production in a different
We knew one of the Japanese transplants in North America very well, of
course, because of IMVP research affiliate John Krafciks tenure there. The New United
Motor Manufacturing Inc. (NUMMI) plant in Fremont, California, is a joint venture
between the classic mass producer, GM, and the classic lean producer, Toyota.
NUMMI uses an old General Motors plant built in the 1960s to assemble GM
cars and pickup trucks for the U.S. West Coast. As GM s market share along the
Pacific Coast slipped steadily, the plant had less and less work. It finally closed its
doors for good in 1982. By 1984 GM had decided that it needed to learn about lean
production from the master. So it convinced Toyota to provide the management for a
reopened plant, which would produce small Toyota-designed passenger cars for the
NUMMI was to make no compromises on lean production. The senior managers
were all from Toyota and quickly implemented an exact copy of the Toyota Production
System. A key action toward this end was the construction of a new stamping plant adjacent to the body-welding area, so that body panels could be stamped in small lots
just as they were needed. By contrast, the old Fremont plant had depended on panels
supplied by rail from GM's centralized stamping plants in the Midwest. There they were
stamped out by the million on dedicated presses.
The United Automobile Workers Union also cooperated to make lean production
possible. Eighty percent of the NUMMI workforce consisted of workers formerly
employed by GM at Fremont. However, in place of the usual union contract with thousands of pages of fine print defining narrow job categories and other job-control issues,
the NUMMI contract provided for only two categories of workers—assemblers and
technicians. The union agreed as well that all its members should work in small teams
to get the job done with the least effort and highest quality.
By the fall of 1986, NUMMI was running full blast. And we were ready to
compare it with Takaoka and Framingham, as shown in Figure 4.2. We found that NUMMI matched Takaoka s quality and nearly matched its
productivity. Space utilization was not as efficient because of the old GM plant's poor
layout. Inventory was also considerably higher than at Takaoka, because almost all the
parts were transported 5,000 miles across the Pacific rather than five or ten miles from
neighboring supplier plants in Toyota City. (Even so, NUMMI was able to run with a
two-day supply of parts, while Framingham needed two weeks' worth.)
It was clear to us by the end of 1986 that Toyota had truly achieved a revolution
in manufacturing, that old mass-production plants could not compete, and that the new
best way— lean production—could be transplanted successfully to new environments,
such as NUMMI. Given these findings, we were hardly surprised by subsequent
events: Takaoka continues to improve, now with much additional automation. NUMMI
is also getting continually better and a second line is being added to assemble Toyota
pickup trucks. Framingham was closed forever in the summer of 1989. SURVEYING THE WORLD
Once we finished our initial survey, we were determined to press ahead on a
survey of the entire world. We were motivated partly by the fact that the companies and
governments sponsoring us wanted to know where they stood and partly by the
knowledge that a survey of three plants could not answer a number of questions about
what roles automation, manufacturability, product variety, and management practices
play in the success of manufacturing.
However, we soon realized that we would have to conceal company and plant
names when we reported our findings. Many companies were willing to grant us access
to their plants only on the condition that we would not reveal plant names in our results.
We have respected their wishes and in this book identify plants only when the company
After four more years of research, we have found the following about
productivity and quality (or accuracy) across the world, as summarized in figures 4.3
and 4.4. These findings are not at all what we had expected. We had anticipated all of
the Japanese firms in Japan to be roughly comparable in performance—that is, equally
lean. Further, we had expected all of the American plants in North America and the American- and European-owned plants in Europe to perform at about the same level
with little variation and to trail the average Japanese plant by about the same degree
that Framingham trailed Takaoka in 1986. Finally, we expected the assembly plants in
developing countries to be marked by low productivity and low quality. The reality is
different. What we find instead is that there is a considerable range ofproductivity
performance in Japan, indeed a difference of two to one between the best plant and
the worst in both productivity and quality. The differences along other dimensions—
space utilization, level of inventories, percentage of the factory devoted to rework
area—are much less, but there is still variation.
In North America, we quickly discovered that Framingham was in fact the worst
American-owned plant. Average Big Three performance in late 1989 was much
better—48 percent more effort and 50 percent more defects, compared with the
Framingham/Takaoka gap in 1986 of nearly twice the effort and three times the
defects. Even more striking, Ford, the originator ofmass production seventy-five years
ago, is now practically as lean in its North American assembly plants as the average
Japanese transplant in North America.5 The best U.S.-owned plants in North America
are now nearly as productive as the average Japanese plant—and are very nearly
equal in quality.
Perhaps most striking was our finding about Europe. Framing-ham, the North
American plant that fared so poorly in comparison with Takaoka and which has now
been closed, in fact had considerably better productivity in 1986 than the average European plant had achieved by 1989. Indeed, as we marched through plant after plant
we came to a remarkable conclusion: Europe, once the cradle of craft production in the
motor industry, is now truly the home of classic mass production. Average American
performance—under unrelenting pressure from the Japanese transplants in North
America—has improveddramatically, partly by closing the worst plants, such as
Framingham, and partly by adopting lean production techniques at others. Europe, by
contrast, has not yet begun to close the competitive gap.
Regarding the Japanese transplants in North America, we found about what we
expected. Their average performance is about comparable to the average Japanese
plant in terms of quality but lags about 25 percent in terms of productivity. We believe
these differences are partly due to the fact that the transplants are still at an early point
in the learning curve with respect to lean production. The differences are also due to
different methods of obtaining supplies that necessitate extra work, a point we will
return to in Chapter 6.6
However, there is important variation among the transplants as well. For
example, one of the transplants has the least efficient utilization of manufacturing
space in the entire world sample. In general, we found that the best-performing
companies in Japan run the best-performing transplants in North America, suggesting
that most of the variation observed is due to differences in management.
Finally, the assembly plants in the developing countries, notably Brazil, Korea,
Mexico, and Taiwan, show an extraordinary range of performance. The best plant in
terms of quality, Ford at Hermosillo, Mexico, in fact had the best assembly-plant quality
in the entire volume plant sample, better than that of the best Japanese plants and the
best North American transplants. The best developing country plant was also
surprisingly efficient, particularly given its modest level of automation. By contrast, the
worst developing country plants were very poor performers, with poor quality and
What accounts for the difference? We believe it can be traced to the assembly
of a product from a lean-development process (as at Hermosillo, where the car
assembled was a variant of a Mazda 323) and doing so with management assistance
from a firm mastering lean production. (In the case of Hermosillo, this was directly from
Ford, but in several other cases an independent firm had received significant and
continuing Japanese management assistance, effectively becoming a transplant.)
These findings require a dramatic reordering of our mental map of the industrial
world, which we believe many readers will find very difficult: We must now stop
equating "Japanese" with"lean" production and "Western" with "mass" production. In
fact, some plants in Japan are not particularly lean, and a number of Japanese-owned
plants in North America are now demonstrating that lean production can be practiced
far away from Japan. At the same time, the best American-owned plants in North
America show that lean production can be implemented fully by Western companies,
and the best plants in the developing countries show that lean production can be
introduced anywhere in the world. THE STRANGE CASE OF THE "CRAFT" PRODUCERS
The productivity and quality data in figures 4.3 and 4.4 are only for mass-market
cars, that is, Fords but not Lincolns, Toyotas but not Lexus, Volkswagens but not
Mercedes. From the outset, we believed that assembly plants are all pretty much the
same in what they actually do, no matter how prestigious the brand they're putting
together. The same type of robots, indeed often identical models from the same
manufacturer, make both the Volkswagen and the Mercedes body welds. Paint is
applied in practically identical paint booths, and final assembly involves the installation,
largely by hand, of thousands of parts as the vehicle moves along a lengthy assembly
line. The real difference between the mass-market and the luxury car is that the latter
may have a thicker gauge of steel in its body, extra coats of paint, thicker insulation,
and many more luxury add-on features.
While obvious to us, this idea is not universally accepted even in the auto
industry and is certainly not the view of the broader public. Repeatedly, executives told
us that our productivity and quality findings might be correct for the average car and
light truck, but "luxury cars are different." We set out to find out for certain by conducting a special world survey of
assembly plants making luxury cars. We went to the Japanese large-car plant that we
believe, based on our survey of the same company's mass-market car plants, to be the
best in the world. In North America, we looked at the Lincoln and Cadillac plants. In
Europe we visited Audi, BMW, Mercedes, Volvo, Rover, Saab, and Jaguar. In each case, wecarefully standardized the tasks being undertaken and the specifications of the
vehicle, so that wewere in fact asking how much effort each plant would need to perform standard assembly steps on the smaller and less elaborate standard car and how
many errors it would make in the process. So the actual amount of effort expended in
each plant is actually much greater than that shown in figures 4.5 and 4.6. In addition,
we adjusted for absenteeism, which runs at 25 percent in many of these European
plants compared with 5 percent or less in Japan. The hours in our table represent
hours actually worked, not hours on the payroll. Our findings were eye-opening. The Japanese plant requires one-half the effort
of the American luxury-car plants, half the effort of the best European plant, a quarter
of the effort of the average European plant, and one-sixth the effort of the worst
Europeanluxury-car producer. At the same time, the Japanese plant greatly exceeds
the quality level of all plants except one in Europe—and this European plant requires
four times the effort of the Japanese plant to assemble a comparable product. No
wonder the Western luxury-car producers are terrified by the arrival of Lexus, Infiniti,
Acura, and the Japanese luxury brands still to come.
In reviewing these data, many readers may wonder if the difference lies in
greater product variety and lower production scale in Europe. Certainly our mental
image of these companies is that of low-volume craft production. In fact, this is simply
not true. The European plants, with one exception, produce at the same volume as the
mass producers we looked at earlier, and in most cases produce a less complex mix of
products than the Japanese luxury-car plant we surveyed.
When we visited the high-quality but low-productivity European plant we just
mentioned, we didn't have to go far to find the basic problem: a widespread conviction among managers and workers that they were craftsmen. At the end of the assembly
linewas an enormous rework and rectification area where armies of technicians in white
laboratory jackets labored to bring the finished vehicles up to the company's fabled
quality standard. We found that a third of the total effort involved in assembly occurred
in this area. In other words, the German plant was expending more effort to fix the
problems it had just created than the Japanese plant required to make a nearly perfect
car the first time.
We politely inquired of these white-smocked workers exactly what they were
doing. "We're craftsmen, proof of our company's dedication to quality," they replied.
These "craftsmen" would have been surprised to learn that they were actually doing the
work of Henry Ford's fitters in 1905—adjusting off-standard parts, fine-tuning parts
designed so as to need adjustment, and rectifying incorrect previous assembly work so
that everything would work properly in the end.
Certainly, these workers are highly skilled and the work they do is no doubt
challenging, since every problem is different. However, from the standpoint of the lean
producer this is pure muda— waste. Its cause: the failure to design easy-to-assemble
parts and failure to track down defects as soon as they are discovered so that they
never recur. When employees don't take this important last step, subsequent assembly
work compounds the initial problem and it's necessary to call for the craftsman to put
Our advice to any company practicing "craftsmanship" of this sort in any
manufacturing activity, automotive or otherwise, is simple and emphatic: Stamp it out.
Institute lean production as quickly as possible and eliminate the need for all
craftsmanship at the source. Otherwise lean competitors will overwhelm you in the
1990s. THE IMVP WORLD ASSEMBLY PLANT SURVEY IN SUMMARY
Figure 4.7 summarizes a number of dimensions of current worldwide
performance of the volume producers at the assembly-plant level in addition to
productivity and quality. In particular, it is striking to note the difference between
average Japanese performance and the average in North America and Europe in terms
of the size of repair area needed, the fraction of workers organized into teams, the
number of suggestions received (and the lack so far of suggestion systems in the
Japanese transplants), and the amount of training given new assembly workers.
One additional and very important finding of the survey bears note: the relation
between productivity and quality. When we first began the surveyand correlated
productivity with quality in all plants, we found almost no relationship. What's more, this
did not change over time. In Figure 4.8, showing the relationship across the world at
the end of 1989, the correlation between productivity and quality is .15. This seemed puzzling. We thought it should either be negatively correlated—
plants with high quality should require more effort to achieve this, as Western factory
managers had long thought—or it should be positively correlated—quality should be
"free," as many writers on Japanese manufacturing had suggested. The answer to the
puzzle, as a moments examination of Figure 4.8 will show, is that both trends are in
evidence and they cancel each other out. The Japanese domestic and transplant
facilities are all concentrated in the lower left corner of the figure. For these lean plants,
quality really is free. Removing these plants leaves a pattern in which plants tend to
have high quality or high productivity but not both. For these mass producers, quality is
expensive when it can be achieved at all. GETTING TO LEAN
We have periodically reviewed our survey findings with practically all the world's
motor-vehicle producers, the main sponsors of the IMVP. So the figures we report here
don't come as a surprise to these companies and are now generally accepted as an
accurate summary of the general state of competition at the factory level.
However, determining who stands where in world competition differs from
explaining precisely what the also-rans need to do to catch up. As we have reviewed our data with these companies, their executives and managers have questioned us on
four points in particular. First, they ask whether automation is the secret. Our answer is that it is and it
isn't. Figure 4.9 shows the relation between the fraction of assembly steps that are
automated—either by robotics or more traditional "hard" automation—and the
productivity of plants. There is clearly a downward slope to the right—more automation
means less effort. (Stated another way, higher levels of automation show a strong
negative correlation (-.67) with higher levels of effort.) We estimate that on average
automation accounts for about one-third of the total difference in productivity between
However, what is truly striking about Figure 4.9 is that at any level of automation
the difference between the most and least efficient plant is enormous. For example, the
least automated Japanese domestic plant in the sample (with 34 percent of all steps
accomplished automatically), which is also the most efficient plant in the world, needs
half the human effort of one comparably automated European plant and a third the
effort of another. Looking farther to the right in Figure 4.9, we can see that the
European plant that is the most automated in the world (with 48 percent of all assembly
steps done by automation) requires 70 percent more effort to perform our standard set
of assembly tasks on our standard car than is needed by the most efficient plant with
only 34 percent automation.
The obvious question is, how can this be? From our survey findings and plant
tours, we've concluded that high-tech plants that are improperly organized end up
adding about as many indirecttechnical and service workers as they remove unskilled
direct workers from manual assembly tasks. What's more, they have a hard time maintaining high yield, because
breakdowns in the complex machinery reduce the fraction of the total operating time
that a plant is actually producing vehicles. From observing advanced robotics
technology in many plants, we've devised the simple axiom that lean organization must
come before high-tech process automation if a company is to gain the full benefit.7
The also-rans' second question is, Does the manufacturability (ease of
assembly) of the product make the difference, rather than the operation of the factory?
Understandably, union leaders have often asked us this question as well. Donald
Ephlin, now retired from his position as vice-president of the United Auto Workers in
the United States, engaged us in a dialogue on this point throughout the life of the
How much of the competitive gap between good firms and bad, he wanted to
know, lies with the unionized workers in the plant and how much with engineers and
managers far away in the corporate development offices. His argument has been
simple: "The workers I represent in American plants are getting the blame for problems
they are helpless to correct." Ephlin argued that putting in place organizational
improvements—just-in-time inventory, a cord that allowed the worker to stop the line,
and so forth—would make a difference, but that none of those improvements could
make a plant fully competitive if the product design was defective.
Answering the manufacturability question definitively is difficult, because we
would need to perform what auto makers call a tear-down analysis on every car being
assembled in every plant we surveyed. Only then could we see how many parts the car
has and how easily they can be assembled. This analysis would be staggeringly
expensive and time-consuming. So we can report only some interesting but partial
evidence that manufacturability is indeed very important. One piece of evidence is a survey we conducted in the spring of 1990 of the
world's auto makers.8We asked them to rank all the other auto makers in terms of how
manufacturable their products are at the assembly plant. They were to base their
ranking on tear-down studies that car companies conduct as part of their competitive
assessment programs. (Strange as it may seem, the first production models of any new
car are unlikely to reach consumers. Instead, competitors buy them, then immediately
tear them apart for competitive assessment.) The results the manufacturers reported
are shown in Figure 4.10. We can't confirm the accuracy of these findings, because we don't know how
much tear-down analysis companies do or how well they do it. When we began our
assembly-plant survey, we were amazed to discover that very few car companies
conducted systematic benchmarking studies of their competitors. Nevertheless, the
companies responding were in close agreement on which producers design the most
manufacturable designs, and the findings correlate nicely with company performance
on our productivity and quality indices. This suggests that manufacturability is
conducive to high performance in the factory.
Further evidence comes from a recent comparison General Motors made
between its new assembly plant at Fairfax, Kansas,which makes the Pontiac Grand
Prix version of its GM-10 model, and Ford's assembly plant for its Taurus and Mercury
Sable models near Atlanta. This comparison was based on tearing down both cars,
then using shop manuals to reconstruct the assembly process.
GM found a large productivity gap between its plant and the Ford plant—both
make cars in the same size class, with similar levels of optional equipment and selling in the same market segment. After careful investigation, GM concluded that 41 percent
of the productivity gap could be traced to the manufacturability of the two designs, as
shown in Figure 4.11. For example, the Ford car has many fewer parts—ten in its front
bumper compared with 100 in the GM Pontiac—and the Ford parts fit together more
easily. (The other major cause of the productivity gap was plant organizational
practices of the type we have just discussed. The GM study found that the level of
automation—which was actually much higher in the GM plant—was not a factor in
explaining the productivity gap.) Ease of manufacture is not an accident. Rather, it's one of the most important
results of a lean-design process. We'll look at this point more carefully in Chapter 5.
A third question that often crops up when we review our survey findings with
companies is product variety and complexity. The factory manager we encountered in
Chapter 3, who maintained he could compete with anyone if he could only focus his
factory on a single standardized product, is typical of many Western managers. This is
certainly an interesting idea and it has a simple logic to commend it.
However, in our survey we could find no correlation at all between the number
of models and body styles being run down a production line and either productivity or
product quality. We tried a different approach by comparing what was being built in
plants around the world in terms of "under-the-skin" complexity. This was a composite
measure composed of the number of main body wire harnesses, exterior paint colors,
and engine/transmission combinations being installed on a production line, plus the
number of different parts being installed and the number of different suppliers to an
assembly plant. The results were even less assuring to those thinking that a focused
factory is the solution to their competitive problems: The plants in our survey with the
highest under-the-skin complexity also had the highest productivity and quality. These
of course were the Japanese plants in Japan.9 LEAN ORGANIZATION AT THE PLANT LEVEL
Those company executives, plant managers, and union leaders accepting our
conclusion that automation and manufacturability are both important to high
performance plants, but that gaining the full potential of either requires superior plant
management, usually raise a final question that we find most interesting: What are the
truly important organizational features of a lean plant—the specific aspects of plant operations that account for up to half of the overall performance difference among
plants across the world? And how can these be introduced?
The truly lean plant has two key organizational features: It transfers the
maximum number of tasks and responsibilities to those workers actually adding value
to the car on the line, and it has in place a system for detecting defects that quickly
traces every problem, once discovered, to its ultimate cause.
This, in turn, means teamwork among line workers and a simple but
comprehensive information display system that makes it possible for everyone in the
plant to respond quickly to problems and to understand the plant's overall situation. In
old-fashioned mass-production plants, managers jealously guard information about
conditions in the plant, thinking this knowledge is the key to their power. In a lean plant,
such as Takaoka, all information— daily production targets, cars produced so far that
day, equipment breakdowns, personnel shortages, overtime requirements, and so
forth—are displayed on andonboards (lighted electronic displays) that are visible from
every work station. Every time anything goes wrong anywhere in the plant, any
employee who knows how to help runs to lend a hand.
So in the end, it is the dynamic work team that emerges as the heart of the lean
factory. Building these efficient teams is not simple. First, workers need to be taught a
wide variety of skills—in fact, all the jobs in their work group so that tasks can be
rotated and workers can fill in for each other. Workers then need to acquire many
additional skills: simple machine repair, quality-checking, housekeeping, and materialsordering. Then they need encouragement to think actively, indeed proactively, so they
can devise solutions before problems become serious.
Our studies of plants trying to adopt lean production reveal that workers
respond only when there exists some sense of reciprocal obligation, a sense that
management actually values skilled workers, will make sacrifices to retain them, and is
willing to delegate responsibility to the team. Merely changing the organization chart to
show "teams" and introducing quality circles to find ways to improve production
processes are unlikely to make much difference.
This simple fact was brought home to us by one of our early studies of Ford and
General Motors plants in the United States. In the Ford plants we found that the basic
union-management contract had not been changed since 1938, when Ford was finally
forced to sign a job control contract with the UAW. Workers continued to have narrow
job assignments and no formal team structure was in place. Yet, as we walked through
plant after plant we observed that teamwork was actually alive and well. Workers were
ignoring the technical details of the contract on a massive scale in order to cooperate
and get the job done.10
By contrast, in a number of General Motors showcase plants we found a new
team contract in place and all the formal apparatus of lean production. Yet, a few
moments' observation revealed that very little teamwork was taking place and that
morale on the plant floor was very low.
How do we account for these seeming contradictions? The answer is simple.
The workers in the Ford plants had great confidence in the operating management, who worked very hard in the early 1980s to understand the principles of lean
production. They also strongly believed that if all employees worked together to get the
job done in the best way the company could protect their jobs. At the GM plants, by
contrast, we found that workers had very little confidence that management knew how
to manage lean production. No wonder, since GMs focus in the early 1980s was on
devising advanced technology to get rid of the workers. The GM workers also had a
fatalistic sense that many plants were doomed anyway. In these circumstances, it is
hardly surprising that a commitment to lean production from the top levels of the
corporation and the union had never translated into progress on the plant floor.
We'll return to the thorny question of how lean production can be introduced into
existing mass-production factories in Chapter 9. IS LEAN PRODUCTION HUMANLY FULFILLING?
As we noted in Chapter 2, Henry Ford’s sword was double-edged. Mass
production made mass consumption possible, while it made factory work barren. Does
lean production restore the satisfaction of work while raising living standards, or is it a
sword even more double-edged than Fords?
Opinions are certainly divided. Two members of the United Automobile Workers
Union in the United States have recently argued that lean production is even worse for
the worker than mass production." They go so far as to label the lean-production
system instituted at NUMMI in California management by stress, because managers
continually try to identify slack in the system— unused work time, excess workers,
excess inventories—and remove them. Critics argue that this approach makes Modern
Times look like a picnic. In Charlie Chaplin's widget factory at least the workers didn't
have to think about what they were doing and try to improve it.
A second critique of lean production comes in the form of what might be called
"neocraftsmanship." This has been put in operation in only a few plants in Sweden, but
it draws wide attention across the world because it appeals to a seemingly unshakable
public faith in craftsmanship.
Let's take Volvo's new Udevalla plant in western Sweden as an example. At
Udevalla, teams of Volvo workers assemble Volvo's 740 and 760 models on stationary
assembly platforms in small work cells. Each team of ten workers is responsible for
putting together an entire vehicle from the point it emerges from the paint oven. Looked
at in one way, this system is a return full circle to Henry Ford's assembly hall of 1903,
which we and the rest of the world left behind in Chapter 2. The cycle time—the interval
before the worker begins to repeat his or her actions—increases at Udevalla to several
hours, from about one minute in a mass- or lean-production assembly plant. In
addition, workers in the assembly team can set their own pace, so long as they
complete four cars each day. They can also rotate jobs within the teams as they desire.
Automated materials-handling delivers the parts needed for eachcar to the work team. Proponents of the Udevalla system argue that it can match the efficiency of leanproduction plants while providing a working environment that is much more humane.
We strongly disagree with both points. We believe that a vital, but often
misunderstood, difference exists between tension and a continuing challenge and
between neocraftsmanship and lean production.
To take the first point, we agree that a properly organized lean-production
system does indeed remove all slack—that's why it's lean. But it also provides workers
with the skills they need to control their work environment and the continuing challenge
of making the work go more smoothly. While the mass-production plant is often filled
with mind-numbing stress, as workers struggle to assemble unmanufacturable products
and have no way to improve their working environment, lean production offers a creative tension in which workers have many ways to address challenges. This creative
tension involved in solving complex problems is precisely what has separated manual
factory work from professional "think" work in the age of mass production.
To make this system work, of course, management must offer its full support to
the factory workforce and, when the auto market slumps, make the sacrifices to ensure
job security that has historically been offered only to valued professionals. It truly is a
system of reciprocal obligation.
What's more, we believe that once lean-production principles are fully instituted,
companies will be able to move rapidly in the 1990s to automate most of the remaining
repetitive tasks in auto assembly—and more. Thus by the end of the century we expect
that lean-assembly plants will be populated almost entirely by highly skilled problem
solvers whose task will be to think continually of ways to make the system run more
smoothly and productively.
The great flaw of neocraftsmanship is that it will never reach this goal, since it
aspires to go in the other direction, back toward an era of handcrafting as an end in
We are very skeptical that this form of organization can ever be as challenging
or fulfilling as lean production. Simply bolting and screwing together a large number of
parts in a long cycle rather than a small number in a short cycle is a very limited vision
of job enrichment. The real satisfaction presumably comes in reyrarking and adjusting
every little part so that it fitsproperly. In/the properly organized lean-production system,
this activity is totally unnecessary.
Finally, the productivity of the Udevalla system is almost certain to be
uncompetitive even with mass production, much less lean production. We have not
audited Udevalla or Kalmar, the two Volvo plants operated on the neocraft model, but
some simple arithmetic suggests that if ten workers require 8 hours simply to assemble
four vehicles (not including welding the body, painting it, and gathering necessary
materials)—for a total of 20 assembly hours per vehicle—Udevalla can hardly hope to
compete with our survey's leading lean-production plant, which requires only 13.3
hours to weld, paint, and assemble a slightly smaller and less elaborate vehicle. Before leaving this point, we offer one final reason why lean production is
unlikely to prove more oppressive than mass production. Simply put, lean production is
fragile. Mass production is designed with buffers everywhere—extra inventory, extra
space, extra workers—in order to make it function. Even when parts don't arrive on
time or many workers call in sick or other workers fail to detect a problem before the
product is mass-produced, the system still runs.
However, to make a lean system with no slack—no safety net— work at all, it is
essential that every worker try very hard. Simply going through the motions of mass
production with one's head down and mind elsewhere quickly leads to disaster with
lean production. So if management fails to lead and the workforce feels that no
reciprocal obligations are in force, it is quite predictable that lean production will revert
to mass production. As one lean-production manager remarked during a plant tour:
"Mass production is simply lean production run by the rule book, so that no one takes
initiative and responsibility to continually improve the system."
This last point raises some profound questions about the spread of lean
production across the whole world, a topic that occupies our attention in Chapter 9.
However, at this point, we need to follow the logic of lean production from the assembly
plant back to product development. As we'll see, the nature of the modern motorcar—a
highly complex product with more than 10,000 parts—requires a highly complex design
and engineering system. And, as in every other aspect of production, the lean
approach to coordinating this system is fundamentally different from that of mass
This was a major change since Satoshi Kamata's slashing critique of working
conditions at Toyota in the early 1970s (Japan in the Passing Lane: An Insider's
Account of Life in a Japanese Auto Factory, New York: Pantheon, 1982 (originally
published in Japan in 1973). In the early 1960s more than 40 percent of Toyota's
workforce was temporary workers without permanent job guarantees. By 1975 all
temporary workers had been converted to permanent workers, a situation that
continued until 1989, as Toyota strained to keep up with the burst of auto demand in
Japan and once more hired workers without permanent guarantees. We will return to
the problems that demand fluctuations create for lean production in Chapter 9.
Throughout the program and throughout this volume we have used information
on product quality provided by J. D. Power and Associates, an American firm
specializing in consumer evaluations of motor vehicles. However, we do not use the
"Power numbers" now routinely cited in automobile advertising in North America.
These numbers are for defects in the entire vehicle. Because we have been interested
in the activities of only one part of the manufacturing system, the assembly plant, we
have obtained data from Power on defects that can be directly attributed to the
activities of the assembly plant. Specifically, these are watjer leaks, loose electrical
connections, paint blemishes, sheet metal damage, misaligned exterior and interior
parts, and squeaks and rattles. Because the Power data are only available for vehicles sold in theUnited States,
the number of European, Japanese, and New Entrant plants for which we can report
quality data is smaller than the number for which we have productivity data and other
indicators of manufacturing performance.
This is the method used by many of the publicly available comparisons of
productivity in auto industry. See, for example, Harbour Associates, A Decade Later:
Competitive Assessment of the North American Automotive Industry, 1979-1989,1989.
For a full explanation of our methods the reader should consult John Kraf-cik, "A
Methodology for Assembly Plant Performance Determination," IMVP Working Paper,
We are pledged not to reveal the identity of specific plants and, by logical
extension, of companies. However, the dramatic improvement in Ford's plant-level
performance in the 1980s is now so well known that it seems unrealistic not to
The advantage of having suppliers across the road who can deliver high-quality
parts directly to the line every hour or two is considerable. In the American transplants,
where most parts are delivered much less fre-quently, considerable effort still goes into
inspecting incoming parts and then transferring them to the point on the line where they
This finding is being borne out by a number of studies in other industries as
well. See, for example, Joseph Tidd, "Next Steps in Assembly Automa¬tion," IMVP
Working Paper, May 1989, for a comparison of recent experi¬ence with automation in
the automotive and electronics industries, and R. Jaikumar, "Post Industrial
Manufacturing," Harvard Business Review, November/December 1986, pp. 69-76, for a
study of flexible automation in machine shops and the watch industry.
For details of this survey see John Krafcik, "The Effect of Design Manufacturability on Productivity and Quality: An Update of the IMVP Assembly Plant Survey,"
IMVP Working Paper, January 1990.
For details on model-mix complexity and under-the-skin complexity as
predictors of assembly plant productivity and quality, see John Krafcik and John Paul
MacDuffie, "Explaining High Performance Manufacturing: The International Automotive
Assembly Plant Study," IMVP Working Paper, May 1989.
We do not mean to suggest that Ford has no plans to eventually renegotiate its
rigid job-control contracts. The contract at the Wayne, Michigan, assembly plant was
recently renegotiated in the direction of a team concept as a prerequisite to Ford's
decision to allocate the new Ford Escort to the plant.
See Mike Parker and Jane Slaughter, "Managing by Stress: The Dark Side of
the Team Concept," in ILR Report, Fall 1988, pp. 19-23, and Parker and Slaughter,
Choosing Sides: Unions and the Team Concept, Boston: South End Press, 1988. ...
View Full Document
- Spring '11