Knowledge, Heavy Manufacturing and Competitiveness: The Case of the Flat Panel Display Industry

Date March 24, 2006
Speaker Thomas MURTHA(Professor of Management, College of Business Administration, University of Illinois at Chicago / Faculty Affiliate, Strategic Management Research Center, University of Minnesota)Stefanie LENWAY(Dean, College of Business Administration and Professor of Management, University of Illinois at Chicago)
Commentator SANBONMATSU Susumu(Consulting Fellow, RIETI / Senior Researcher, Organization for Small & Medium Enterprises and Regional Innovation, JAPAN / Visiting Professor, Hitotsubashi University)
Moderator ASAKAWA Kazuhiro(Faculty Fellow, RIETI / Professor, Graduate School of Business Administration, Keio University)


In this presentation, Professors Thomas Murtha and Stefanie Lenway discussed a research project funded by the Alfred P. Sloan Foundation. The first part of their talk was based on a book that they published in January 2002 called Managing New Industry Creation, which they co-authored with Jeff Hart. The book discusses the business history and corporate strategies that made possible the global flat panel display (FPD) industry, especially the TFT LCD sector. In 1995 the Alfred P. Sloan Foundation was searching for a team to study the industry, to address many concerns that existed in the United States because there were no producers who were setting up high volume, large-format manufacturing facilities there. Professors Murtha and Lenway were asked to evaluate policies that were being implemented at the time. Their discoveries were quite surprising. After their book was published, they were invited by the Foundation to study further what they had discovered about the TFT LCD industry in order to see if those findings applied in other industries. They found that global collaboration among companies from many countries was critical to the industry's emergence, progress and continuing innovation.

The Sloan Foundation is an independent foundation in the U.S. that sponsors a variety of programs, including Industry Studies. The goal of Industry Studies is to encourage academics to interact with managers in order to create a deep understanding of real industry phenomena as well as to build new theories that can form the basis for practical recommendations and learning. The Foundation often describes its role as "fostering understanding of the role of technology in economic growth." There are more than 22 Sloan industry study centers, as well as many smaller projects around the U.S. The Sloan Foundation provided generous funding for the study of FPDs. The new project, funded in 2003, is called "Creating and Capturing Value in Global Knowledge Networks." It brings together six of these centers and projects. Professors Murtha and Lenway are the principal investigators for the FPD team. RIETI Faculty Fellow, Professor Kazuhiro Asakawa of Keio University, is also a member. Professor Murtha also serves as team co-coordinator for the overall project.

The "Industries as Global Knowledge Networks" project seeks to understand the role of inter-organizational or inter-firm collaboration in innovation. Innovation keeps companies moving forward, makes them unique and helps them sustain competitiveness. This is true whether a company competes on cost or on differentiation. Companies need to keep costs down, innovate in new product development, manufacturing processes, sourcing and in business management models. New, unique products, such as the liquid crystal television, can bestow tremendous competitive advantage on a company if it can succeed in manufacturing at a price the market can afford. It is often surprising when companies cooperate on such advanced technologies, because the theory of the firm suggests that companies prefer to innovate independently and keep their discoveries as secret from each other as possible. The idea of collaboration for innovation creates a number of management challenges for firms and for government policy. These challenges are the focus of this study.

There are many interesting new phenomena that involve inter-firm collaboration in the world today. Many companies that were vertically integrated have separated into smaller companies that still work together. Contract manufacturers are producing electronics goods for a wide range of companies; semiconductor foundries take designs from many companies and segregate all of their customers' private information within their facilities while providing a general manufacturing service. Outsourcing of business processes has created many opportunities for cost savings in places like India, a country that, at the same time, has become a hotbed of innovation. Then there is the whole question of whether companies need to take advantage of opportunities such as business process outsourcing just to maintain efficiency in their innovation processes since a project requiring many trials has a better chance of success if the costs of trials can be lowered so that their number can be increased.

In Japan in 1995 there were many facilities that produced FPDs or produced materials and equipment for making them. However, several of the most important among these were companies from other places in the world that came together in Japan to create the new, high volume, large-format FPD industry. The true beginning for the industry can be traced to when the first high-volume, large-format TFT LCD displays were produced, which dates back to about 1990 and NEC. Much history and research and development took place before that, but in the early 1990s, when the high-volume, large format manufacturing process paradigm was being created, a number of companies from around the world - including Applied Materials, IBM and Corning (U.S.), and Merck (Germany) - worked through their Japanese affiliates, together with Japanese companies, to help make that happen. The bets that they made in 1990 have paid off rather handsomely. In a recent meeting between the researchers and senior managers at Nikon, one manager suggested that the industry was born in 2000 because that was the first time five companies made a profit at the same time. Even if that is so, different companies have made profits at different times, and it has been very challenging for the industry because of pressure to push costs and prices down in order to create the mass market and achieve the goal of commercializing large, flat televisions for the mass market.

Given that the industry established itself in Japan in the early 1990s and that country held about 95% of market share until the late 1990s, it is surprising the degree to which the leadership in FPDs has changed hands over the last five or so years. We are left with many questions. In particular, why did the industry emerge with its geographic center in Japan? Why did the industry diffuse to other Asian countries? Why not the U.S.? What role should Japan play in the industry's future? Debate over these concerns in Japan seems very much like debate that took place in the U.S. in the 1990s, when there was almost no local production. The U.S. government tried to create a program offering financial incentives for companies to get started in the business on the condition that they work together. No incentives were offered to companies from other countries. A key finding in the presenters' book reveals that none of the producers, equipment makers or material suppliers that accepted the incentives and worked exclusively together were successful. The previously mentioned successful companies, however, did not accept incentives from the government nor try to concentrate partner, customer and supplier relations in the U.S. They went to Japan and operated through their strong Japanese affiliates. All are successful today. Very few major equipment suppliers exist in any given segment of the TFT LCD value chain. In order to succeed in this business, an equipment supplier typically must have a very substantial market share. Canon and Nikon pretty much divide the market for lithographic tools, and such U.S. companies as AKT (chemical vapor deposition), Corning (glass substrates) and Photon Dynamics (test equipment) enjoy similar high levels of success.

Research for this project included an examination of media, company documents and market research data, but primarily involved talking to managers in order to understand the thought processes that go into the strategies that make it possible to compete in the industry. Between 1996 and 2006 the project team visited many sites around the world and discovered four basic factors required for a company to participate in high technology industry emergence: continuity, learning, knowledge network access for collaboration, and the ability to move quickly. Those successful companies that followed Japan shared an additional factor: "stepping back to learn to step forward."

In 1968 RCA demonstrated the first LCD. Shortly thereafter, Seiko and Sharp research projects began with the goal of creating clock and calculator displays. At the same time IBM was making plasma (PDP) teller screens for automated cash machines. Progress was slow until 1983, when Seiko demonstrated an LCD color TV. Although it did not go far commercially, it alerted everyone to the technological and business potential of the technology. IBM moved to studying large-format LCD in partnership with Toshiba. By 1988 both IBM/Toshiba and Sharp had produced 14-inch prototypes. However, manufacturing, which started for both companies in 1991, was very low-yield, and by 1993 it was clear that significant advances needed to be made in manufacturing production technology. A turning point in yield improvement came when Applied Materials became a supplier to the industry, establishing AKT to introduce chemical vapor deposition built on semiconductor technology. In 1995 Sharp, then the IBM-Toshiba joint venture DTI, started third-generation lines, moving to larger substrates. After 1996 the Asian financial crisis retarded investment in third-generation lines in Japan. The next line went up at Samsung, who shared and then seized the large-format market share lead from DTI after 1998. In 1999 Taiwanese companies entered the market, and by 2001 Taiwan charted a major company among the market share leaders. The first year that CRTs were less valuable in the market than flat panel televisions was 2003. Since 2004, Taiwan and Korea have been neck and neck for global market share leadership. The share in Japan has diminished substantially. IBM left production at about the same time as other Japanese companies cut back. The successful companies started small, built up incrementally over time, and were able to achieve participation in large format. In summary, collaboration and the global knowledge network played key roles in success: successful companies from outside Japan came to Japan in order to partner.

The kind of continuous learning that has taken place in this industry is most evidenced by the progressive mastery of technological challenges presented by ever-increasing substrate sizes across generations and within generations. According to some, this is one of the fastest changing industries, with 18 times the pace of change compared to an equivalent period of time in the earliest days of the semiconductor industry, and it still appears to change at double the pace of Moore's Law. Successful companies are those that have been able to stay abreast of this.

"Stepping back to leap forward" pertains to Taiwan and Korea. When the industry was taking off in Japan, companies in the U.S. were still thinking about it, believing that they should enter with a more advanced generation. That proved impossible, because the industry was changing so rapidly that much key knowledge for progress at the frontier of generational change existed in peoples' minds rather than on paper or within equipment and materials. Business managers needed to share that knowledge directly in order for next generation manufacturing process innovation to happen. Companies wishing to lead generational shifts tend to accomplish it by moving their experienced operators and engineers forward to start up the new lines. Companies without experienced people are unable to enter this business unless they step back. That is what Taiwan and Korea did, starting with generation one or two plants at the time of generation two or three, thus establishing a knowledge foundation on which they could move forward. For the older platforms, a great deal of technological uncertainty has been wrung out of the system. China would like to enter at generation six, which is where things mainly are right now. Generation-seven plants have started up but with difficulty. Generation eight is a year or so away. China has not learned to step back.

To summarize, the FPD sector is a knowledge-driven industry based on a manufacturing paradigm that emerged in the mid-1990s from global collaboration in Japan. It has been fundamentally scalable to larger and larger substrates over time. The main question now is, when will this rapid change come to an end? Intense interpersonal contact among and within companies remains critical because rapid technology evolution creates a knowledge codification backlog.

Why did the industry start in Japan? Why was it necessary for companies who wanted to be part of the industry to be in Japan? The answer can be found in the notion of a knowledge codification backlog. The faster an industry is changing, the greater the proportion of knowledge in that industry that remains in people's heads or in verbal tradition. In addition, with the industry emerging in Japan, in the earliest days, translations of important documents and media reports often lagged one year behind. It therefore became more important for people working in the industry to be concentrated in particular geographic areas in order to share knowledge. This is a critical factor.

The industry still shares many of these characteristics, although it has diffused somewhat. Korean companies were quite aggressive and purchased the "global toolset," hired many Japanese consultants to come and help them, and they stepped back in order to leap forward. This explanation applies even more closely to Taiwan. Japanese companies aggressively partnered with Taiwan to create second sources.

The only companies that lead generational changes today are Korean and Japanese: Samsung, LG. Philips and Sharp. Companies that adopt follower strategies are dealing with a more highly codified knowledge set because they can buy the same equipment and materials as were built for the leaders. But they nonetheless must add a good deal of process knowledge and creativity in order to start up and run a new generation plant. They deal to some extent in a modular setting but it is not plug and play, it is plug and pray. At the generational leading edge it is neither plug nor play because nobody knows how new equipment is going to work until they start it up. It takes a lot of entrepreneurial courage to be in this industry because every generation presents a new set of technical challenges. In contrast, older generations are well settled down, equipment is available on the market and experienced people can be hired. So companies such as those in Taiwan that adopt strategies based on following generations tend to wait a year or two, which diminishes technical uncertainty. However, they still have to create the manufacturing process. Countries like China that want to get into the business are trying with third or fourth generations, but would like to move to six or higher. The Taiwanese are in somewhat of a plug and pray mode and the Chinese are definitely in that mode, but perhaps not praying enough so far.

This once small-scale industry has turned into heavy industry; it is extremely expensive, the equipment is gigantic, and that creates a whole set of collaborative and competitive dilemmas. Knowledge in the industry tends to exist in concentrated geographic areas and now that the plants have become so huge, companies need to build in proximity to one another. There will potentially be a continuous flow of raw materials. Suppliers know a lot in this business. Some people will say that the suppliers know more about starting up fabs and bringing them up to commercial yield than the producers do. The way the industry is organized there is still a need for global interaction among equipment and material suppliers for those companies who want to remain in the technological lead. Furthermore, those suppliers need to have access to those tools and equipment in operation in order to follow up. This creates an uncommonly collaborative industry, particularly for those companies determined to be on the leading edge of generational change; when starting up a new plant they must invite everybody involved in the process to come into their plants. That means that the process improvements and innovations and new knowledge that must be created to integrate the new manufacturing lines are potentially being widely shared across companies within these fabs. The kinds of tweaks that manufacturers like to make in their manufacturing equipment and material sets in order to differentiate themselves could easily become known at least to the equipment manufacturing vendors or the materials makers who provide that particular piece of equipment. It is a strange industry that generalizes so much knowledge through a shared supplier base; it is less vertically integrated than many such industries in the past.

It is sometimes said that the biggest competitive secret to be able to operate well in the industry is the ability to keep secrets. There are managerial processes, relationships of trust, understandings and techniques that go into maintaining some semblance of privacy, of proprietary secrets so that companies can compete on differentiation. At the same time there exists an understanding of the need for a shared knowledge base for industry progress, and none of the producers want to pay for the R&D that is taking place among the equipment and material makers. Manufacturers would much rather share responsibility widely within the knowledge network.

There seems to be a certain amount of hubris in those companies that talk about vertical integration as a goal. The only way to vertically integrate is to start to buy the equipment and materials suppliers. The estimates are 5-10 years to create sufficient knowledge within the company to be able to even make an entry in a particular equipment category. Considering that a new generation is coming every one and a half years or so, how could a company even decide a target an entry time if they want to be in that business in five or 10 years? What is known by the people who work in the equipment and material suppliers is vital. This is how producers leverage their own work forces. If new fabs require the same experienced people to start up, so how can these companies ever expand their capacity? In the early days it was only through robotization. Continuing generational shift in many companies took place in an atmosphere of stable headcount. That is no longer true. The people who come on to operate plants after they are started need to be heavily trained, and it is these same teams that go forward to start up the new generation. The knowledge of the people working in equipment and materials is a critical leverage factor.

Not all knowledge can be codified; there is always some part that is personal. But the process of knowledge creation in an organization is a process of socialization; of converting personal knowledge to explicit knowledge that can be shared and replicated easily over time, as yields rise. In building a new fab, a large equipment supplier is going to come in with an installation team that is entirely built of its own personnel - knowledge carriers who are usually augmenting some local team. They get the tool started, and only then does the operator force come in. But the install team will remain for some time as the other companies put up their tools in order to perform line integration. Even as yield is improving, equipment makers' teams may return to help with experiments to create further improvements. So, TFT LCD producers may still be running experiments with major assistance from resident teams from outside the firm, even after months have passed. This is a model of collaboration that is quite challenging. What are the techniques are for making this work? Companies say they never talk to their clients about what they can and cannot talk about. If they started talking about this, the conclusion would be to forbid talking at all. So there is an immense amount of trust and mutual understanding that exists within this industry as a social community itself.

It is quite possible that the industry is coming to a crisis of confidence within the next couple of years. Across seven generations, the industry has experienced increasing returns to scale in manufacturing. The laws of economics suggest that at some point, a generation will come when plants experience decreasing returns to scale. It is reasonable to think that at some point a plant will be put up that is inefficient. The potential is there for a very expensive mistake. If that were to happen, conservatism could set in, producing a slowing or halt of generational advance. Under those circumstances the level of collaboration seen today could change as the basis of competition becomes entirely manufacturing process improvements on a stable toolset, and there will not be the advantage of introducing new generational tools. When, not whether this will happen is a critical question facing the industry today. In these circumstances, the major question facing the industry must be "what new manufacturing paradigm will replace the present one?

Following are comments from SANBONMATSU Susumu
Consulting Fellow, RIETI / Senior Researcher, Organization for Small & Medium Enterprises and Regional Innovation, JAPAN / Visiting Professor, Hitotsubashi University

I understood the knowledge creation process for exploiting the new industry generation in East Asia, so I have some functional ideas about the growth of the industry. Also, regarding the global teams, we are now deciding to make a Japanese team to analyze or understand regional innovation for the LCD-making process. It is now in process. Professor Asakawa is largely committed in those global teams, so if possible we Japanese teams will make a new contribution in this area.

In addition, I would like to raise a few questions: how has the increase in panel size influenced the forward progress in generations, in terms of motherglass size? What impact has the increase in panel sizes had on companies' decisions to continue to invest in new manufacturing equipment designs?

Response from Professor Stefanie LENWAY
Dean, College of Business Administration and Professor of Management, University of Illinois at Chicago

The relationship between panel size and the LCD business cycle or crystal cycle has been a very interesting one. To understand it, it is useful to go back to 1996, when the industry was moving from generation-two, 10.4-inch to generation-three, 12.1-inch screens. The question for suppliers and manufacturers is "what are consumers going to buy?" Once the 10.4-inch display was proven and the laptop computer business was born everyone went into the market, so there was a tremendous overcapacity. Nevertheless DTI and Sharp put up plants optimized for 12.1-inch displays, and they were such a hit that all the companies with generation-two lines started making 12.1-inch displays. But production capacity was cut in half for companies that made the larger displays on the older substrate sizes. The prices went up, the market stabilized and there was incentive for companies to build more fabs. This is the crystal cycle, a dynamic between increased prices due to shortage creating incentives for large-scale investment, and then overcapacity where the prices go down, expanding demand tremendously. This puts price pressures back on the product and you get a stabilization. However, even in American houses, when you get to 40 inches there is not room for a bigger television so it becomes throughput efficiencies that drive the advances in generations rather than final product markets.

Do these clusters of production like Kameyama and Gumi create new, better knowledge and more efficiency? CMO is on its own and may be the most innovative display company in the world. I do not think we have all the answers.

Studying regional innovation, we keep going back to that map of Japan and saying that it is a global industry. Even though you see production in Asia you see knowledge creation globally, and if you miss that you miss the key point of the story.

Questions and Answers

Q: Where does cooperation stop and competition begin between suppliers and producers, and horizontally between producers themselves, in the LCD industry? Also, at some point somebody needs to lead innovation; who exactly is that leader?

A: This industry was created partly because the producers were very open with the equipment makers. There was a real sense at the beginning that there was a cost of leadership in terms of generating knowledge and diffusing it and it was worth it. It is a dynamic of the industry and companies have to make a strategic decision. The leadership and investment are very elusive, I think.

Process innovation is one kind of leadership. Leadership has been defined in this industry as who gets up the first next-generation plant. When things were changing rapidly it did not matter as much that things were diffusing rapidly. The cost advantage came from being ahead of the cyclical price declines that took place in the industry, which they caused to try to create a mass market. I think that the definition of leadership will have to shift if we see a slowing of generational progress.

Q: Who defines standards in this industry? Is there any kind of standard or platform?

A: There are no standards, at least in terms of substrate size which drive everything else. But at some point it will get so big that they will start using one platform.

Q: Is there cooperation in defining the systems or the standards in some way?

A: The two challenges are cost and yield. They want to make displays bigger because they are cheaper, but also to make them with no defects. Much of the knowledge created and diffused was over process and materials improvements that reduce display defects.

Q: Why is this industry impossible for people in the U.S. to engage in?

A: This is a knowledge-driven industry. When Sharp started producing displays with RCA's license the knowledge shifted to Japan. We have seen the industry move to Korea through the toolset and to Taiwan through licensing. The key to this industry is to adopt strategies to hook into the knowledge. American equipment and materials suppliers have been very successful but nobody is interested in making panels themselves.

Q: Why is there no interest?

A: If we do see production in the U.S. an operator force will have to be brought in from elsewhere for at least a time. But most of the display projects in the U.S. are on alternative technologies and potential U.S. producers have remained obsessed with making a breakthrough, of leapfrogging over billions of dollars of investment and installed capital base.

Q: Is Taiwan's position as a close following generation-based producer a valuable positioning; are they happy to be a follower? How is China relative to that?

A: Taiwan is seen as a follower because its companies are newer, and that is their strategy. They went in with generation four and five. They are now building generation-six fabs, but somewhat behind the leaders in Japan and Korea. But they are making a lot of process innovations which will ultimately catapult them into a leadership position. They will jump the strait and end up on the mainland, but not with a new company jumping into generation six. Some companies are setting up production in Taiwan just to learn how to do it and prepare to go to China.

Taiwanese companies are happy to be followers at the beginning; that is their explicit strategy. They will build generation-seven fabs, too, but they have not done so as yet because nobody else has completely solved the difficulties of consistently running these plants at high yield. They will not put up generation eight before anyone else does unless they change their strategy.

Q: Does this have any connection with their capacity in research and development?

A: No. Their strategy is explicitly to come in later as the technological uncertainty surrounding a given generation diminishes so as to not spend so much money on wasted materials while solving the early yield problems. Then by doing process improvements on the existing generation they believe their margins will be equal to those of the Japanese and Koreans. It appears that they are correct about that. Their big problem is that they cannot get any brands going.

Q: Does the Taiwan government have a large role to play in the second-mover strategy?

A: Taiwan was very clever. It told the Taiwanese in the U.S. to come back and do whatever experiments it wanted at ITRI's R&D factory. Then it went on to ACER and then UNIPAC, which merged and put up a generation four fab, as well as other companies. Also Taiwanese can work at government labs as an alternative to military service. That has helped, as shortages of people are one of the biggest constraints on industry development. Taiwan also helps a lot with real estate, which is about impossible to obtain otherwise because there is so little room.

*This summary was compiled by RIETI Editorial staff.