THE COMPETITIVE SEMICONDUCTOR MANUFACTURING HUMAN
Second Interim Report
Clair Brown, Editor
11. Managing Creativity and
Control in Innovation
11.2 The Development Process
11.3 The Development Process at USTech
11.4 The Development Process at JapanTech
11.5 Engineering Innovations on the Shop
This study focuses on the management of innovation in the semiconductor
industry, where innovation is critical to both short-run and long-run
competitiveness. In particular, we examine how work organization,
incentive systems, and communication systems affect the creation,
sharing, and control of knowledge. The semiconductor industry is
characterized by rapid technological change, high capital costs,
continual price declines, and strict quality standards. These industry
characteristics result in high risks and returns to product innovation
and in competitive pressures for continual reduction of manufacturing
costs. They also make the semiconductor industry ideally situated
for exploring technological innovation and knowledge diffusion.
Innovation can be divided into two fairly distinct activities in
terms of when and where the innovation occurs. One type of innovation
occurs in the development of a new product or process, and the second
type occurs in improvements on the shop floor of a product or process.
Long-run competitiveness of a leading-edge company depends on its
product line, while short-run competitiveness depends on price and
reliability. Over time, a leading-edge company must develop new
products that the market values and must manufacture a quality product
in a cost-effective fashion. The challenge for an industry leader
is to bring a new product to market and recoup development costs
during the short period when competition is limited and prices are
relatively high. An industry follower must enter the market with
the comparable product at a low cost, since the entry of competitors
(followers) into the market causes prices to drop rapidly. This
type of price competition drives shopfloor innovation, which determines
the cost and quality of the product. Both long-run and short-run
innovation processes are critical to company performance, but their
relative importance to the firm depends upon the companys
This study focuses on innovation by engineers in two specific activities--developing
new products or processes and transferring the process to high volume
fabrication. The management of tension between control and creativity,
which are inherent in innovative activities, is studied by documenting
and analyzing examples of how these tensions are successfully managed
at a leading Japanese producer of memory (pseudonym "JapanTech")
and a leading American producer of logic (pseudonym "USTech").
In development activities, two major tensions are studied:
the tension between encouraging new
ideas and the need to choose only a few of the ideas to be used
in the new product or process;
the tension between encouraging individuals
to own and develop new ideas and encouraging engineers to share
their ideas with a team for more rapid evaluation and development
of the idea.
On the shop floor during the transfer, two major tensions are studied:
preventing unauthorized improvements
by engineers while keeping them actively involved in solving
problems and suggesting improvements;
having manufacturing engineers remain
engaged in less interesting, but necessary, tasks such as documentation
and problem solving rather than doing the more challenging tasks
such as process improvements, which are not part of their assignment.
Of the engineers working in development and fabrication, only a
subset are involved in both the development and transfer of a particular
new process. Deciding the composition of this subgroup is one part
of setting up the employment system that structures and encourages
The development process is somewhat shaped by the type of device
being created. The development strategy is also affected by the
product market. Producers of logic devices, which are distinguished
by their architecture, must maintain their lead in developing improved
capabilities. Producers of memory devices, which are a commodity
product without distinguishing characteristics within a generation,
must keep up with their competitors in unveiling the next generation
and in producing high quality, low cost products. Development of
memory devices is process driven; development of logic devices is
driven by the architecture. Although development of memory devices
used to be the technology driver for all companies, now issues specific
to logic devices such as multi-level interconnect have become the
technology drivers for logic producers. Nishi describes the development
process as top-down for logic integrated circuits, which proceeds
from the system architecture to a logic diagram to a circuit design,
and as bottom-up for memory integrated circuits, which proceeds
from a basic understanding of science to development of new materials
and new processing techniques to new device structure. Nishi argues
that the careers of those people who worked on the technologies
that were not used continue to develop the technology for the next
generation and so their careers do not suffer. We observed that
sometimes this occurs. Okimoto and Nishi argue that Japanese practices
advance research development in memory technology, which has predictable,
linear trajectories, but Japanese practices constrain advances in
logic technology, which has nonlinear, highly volatile trajectories.
Transfer problems also differ in that memory devices must control
leakages and logic devices have problems with interconnects and
are harder to test.
Although comparisons across memory and logic producers impose constraints
on employment structures, we have observed both differences and
similarities in the management of technology regardless of the type
of device. The two companies studied are successful in the innovation
process both in development and in high volume manufacturing. This
research is based upon several field work trips to USTech and JapanTech.
In-depth interviews with over a dozen engineers and managers at
each company were conducted. These interviews are supplemented by
a survey of engineers from a wider range of companies regarding
methods engineers use to collect and process information and how
their work is structured and rewarded.
As we will see, the management of creativity and control operates
differently in Japan compared to the US. Since large firms in Japan
have similar compensation systems, which are largely shaped for
the junior engineers through negotiations with the union in national
annual wage bargaining and through widespread national employment
practices, much of what we observed at JapanTech is representative
of other large Japanese electronics companies. In the U.S., few
electronics companies have any union representation, and employment
practices are more widely varied across companies than in Japan.
Therefore, USTechs employment system cannot necessarily be
thought of as representative of other large U.S. semiconductor companies.
We will indicate which practices are widespread and which seem to
be idiosyncratic in our analysis of USTech.
JapanTechs HR system relies on teamwork and group responsibility
with engineers engaging in a broad range of job tasks. Their education
is developed through company-based classes, mentoring, and job rotation.
Compensation reflects specific career ladders rather than company
or individual performance. Engineers participate in knowledge sharing
through presentations at conferences and patent applications.
In contrast, USTechs engineers specialize in development (advanced
degrees) and fabrication activities (BS degree). Manufacturing engineers
are prohibited from making unauthorized improvements in the process.
The development engineers are given individual autonomy and responsibility
with large monetary rewards for good performance. USTechs
policy is not to share knowledge with the outside world; engineers
rarely write or present papers or submit patents (except defensively).
USTechs practices reflect educational requirements in development
and fabrication, the leading edge nature of the technology under
development, and the difficulty in qualifying a logic chips
characteristics after process modifications. Compared to USTech,
other U.S. semiconductor companies have less strict policies of
knowledge sharing, since they usually believe they have something
to learn in the process. Also, other companies usually allow more
improvements during the transfer, since the process is not as well-developed
at the hand-off. Other companies usually do not give as much responsibility
to new graduates; however, giving engineers individual autonomy
and performance rewards are widespread practices in the U.S.
Organization of development and transfer activities both accommodates
and requires these differences in the U.S. and Japanese HR systems,
and simultaneously they determine innovation, diffusion, and control
of knowledge. Precisely those structures of the Japanese firm that
support team-based learning and problem-solving impose constraints
on individual initiative and autonomy; for example, a system that
supports knowledge sharing through public channels and supplier
relationships raises potential problems of control of knowledge.
Precisely those structures of the American firm that support individual
creativity and breakthroughs impose problems of control over the
process; for example, a system that restricts knowledge sharing
raises potential inefficiencies associated with duplicating knowledge
or developing inferior solutions.
A trade-off appears to exist between supporting information sharing
for joint problem-solving and supporting individual creativity in
problem-solving. The Japanese human resource system has highly developed
systems to support interfirm knowledge creation and sharing (i.e.,
the joint development of ideas with or acquired knowledge from other
firms) and intrafirm knowledge creation and sharing (i.e., the joint
sharing of knowledge and skills among employees within a team and
across groups). The U.S. human resource system is better at structuring
and rewarding individual, as opposed to group, initiative and endeavors.
Although U.S. companies do not have a history of interfirm knowledge
sharing, American semiconductor companies recently have been experimenting
with joint ventures with other companies, largely in response to
the extremely high capital and research and development costs.
Go to Next Section of this Chapter
Go to Table of Contents for this
Go to Table of Contents for the
CSM-HR Interim Report