Technology and Global Industry: Companies and Nations in the World Economy (1987)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 65 STRATEGIC PARTNERING, AND LICENSING DECISIONS Capturing Value From Technological Innovation: Integration, Strategic Partnering, And Licensing Decisions DAVID J. TEECE Why, and under what circumstances, is the recognized technological progressiveness of a nation not sufficient to capture the benefits stemming from its capabilities in science and technology? This chapter examines why firms and nations can lose ground in the commercialization of advanced technologies at a time when they are the principal sources for major technological innovations of industrial significance; the capacity for scientific and technological innovation may be the last rather than the first advantage that a mature economy loses as it enters its declining phase. The framework developed here helps identify the factors that determine who wins from innovation: The firm that is first to market, follower firms, or firms that have related capabilities that the innovator needs. The follower firms may or may not be imitators in the narrow sense of the term, although they sometimes are. The framework helps to explain the share of the profits from innovation accruing to the innovating firms and nations compared to its followers and suppliers. THE PHENOMENON A classic example of the phenomenon considered in this chapter is the computerized axial tomographic (CAT) scanner developed by the U.K. firm Electrical Musical Industries (EMI) Ltd.1 By the early 1970s, EMI This chapter is a revised version of a previously published paper by David J. Teece: "Profiting from Technological Innovation," Research Policy. Vol. 15 (1986), No. 6. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 66 STRATEGIC PARTNERING, AND LICENSING DECISIONS was in a variety of product lines, including phonograph records, movies, and advanced electronics. EMI had developed high-resolution televisions in the 1930s, pioneered airborne radar during World War II, and developed the United Kingdom's first all solid-state computers in 1952. In the late 1960s, the pattern recognition research of Godfrey N. Hounsfield, an EMI senior research engineer, resulted in his being able to display a scan of a pig's brain. Subsequent clinical work established that computerized axial tomography was viable for generating cross-sectional 'Mews" of the human body, the greatest advance in radiology since the discovery of x rays in 1895. Although EMI was initially successful with its CAT scanner, within 6 years of its introduction into the United States in 1973, the company had lost market leadership and by the eighth year had dropped out of the CAT scanner business. Other companies successfully dominated the market, though they were late entrants, and are still profiting in the business today. A further example is that of the Royal Crown Companies, Inc., a small beverage company that was the first to introduce cola in a can and the first to introduce diet cola. Both Coca-Cola and Pepsi-Cola followed almost immediately and deprived Royal Crown of any significant advantage from its innovation. Bowmar Instrument Corporation, which introduced the Figure 1 Taxonomy of outcomes from the innovation process 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 67 STRATEGIC PARTNERING, AND LICENSING DECISIONS pocket calculator, was not able to withstand competition from Texas Instruments, Hewlett-Packard, and others, and went out of business. Xerox Corporation failed to succeed with its entry into the office computer business, even though Apple Computer, Inc., succeeded with the Macintosh, which contained many of Xerox's key product ideas, such as the mouse and icons. The story of the DeHavilland Comet has some of the same features. The Comet I jet was introduced into the commercial airline business 2 years or so before Boeing introduced the 707, but DeHavilland failed to capitalize on its substantial early advantage. MITS introduced the first personal computer, the Altair, experienced a burst of sales, then slid quietly into oblivion. If there are innovators who lose, there must be followers (imitators) who win. A classic example is IBM Corporation with its PC, a great success from the time it was introduced in 1981. Neither the architecture nor the components of the IBM PC were considered advanced when introduced; nor was the way the technology was packaged a significant departure from the then-current practice. Yet the IBM PC was fabulously successful and established MS-DOS as the leading operating system for 16-bit PCs. By the end of 1984, IBM had shipped more than 500,000 PCs and may have irreversibly eclipsed Apple in the PC industry. Figure 1 presents a simplified taxonomy—with examples—of the possible outcomes from innovation. Quadrant 1 represents positive outcomes for the innovator. A first-to-market advantage is translated into a sustained competitive advantage that either creates a new earnings stream or enhances an existing one. Quadrant 4 and its corollary quadrant 2 are the focus of this paper. PROFITING FROM INNOVATION: BASIC BUILDING BLOCKS To develop a coherent framework within which to explain the distribution of outcomes illustrated in Figure 1, three fundamental building blocks must be put in place: the appropriability regime, the dominant design paradigm, and complementary assets. Regimes Of Appropriability A regime of appropriability refers to the environmental factors, excluding firm and market structure, that govern an innovator's ability to capture the profits generated by an innovation. The most important dimensions of such a regime are the nature of the technology and the efficacy of legal mechanisms of protection. It has long been known that patents do not work in practice as they do 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 68 STRATEGIC PARTNERING, AND LICENSING DECISIONS in theory. Rarely, if ever, do patents confer perfect appropriability, although they do afford considerable protection on new chemical products and rather simple mechanical inventions. Many patents can be "invented around" at modest costs. They are especially ineffective at protecting process innovations. Often patents provide little protection, because the legal requirements for upholding their validity or for proving their infringement are high. In some industries, particularly where the innovation is embedded in processes, wade secrets are a viable alternative to patents. Protection of trade secrets is possible, however, only if a firm can put its product before the public and still keep the underlying technology secret. Usually only chemical formulas and industrial-commercial processes (for example, cosmetics and recipes) can be protected as trade secrets after they are placed on the market. The degree to which knowledge is tacit or codified also affects ease of imitation. Codified knowledge is easier to transmit and receive and is therefore more exposed to industrial espionage and the like. Tacit knowledge by definition is not articulated, and transfer is hard unless those who possess the know-how in question can demonstrate it to others (Teece, 1981). Survey research indicates that methods of appropriability vary markedly across industries, and probably within industries as well (Levin et al., 1984). The property rights environment within which a firm operates can thus be classified according to the nature of the technology and the efficacy of the legal system to assign and protect intellectual property. Though a gross simplification, a dichotomy can be drawn between environments in which the appropriability regime is "tight" (technology is relatively easy to protect) and "loose" (technology is almost impossible to protect. Examples of the former include the formula for Coca-Cola syrup; an example of the latter is the Simplex algorithm in linear programming. The Dominant Design Paradigm Two stages are commonly recognized in the evolutionary development of a given branch of a science: the pre-paradigmatic stage when there is no single, generally accepted conceptual treatment of the phenomenon in a field of study, and the paradigmatic stage, which begins when a body of theory appears to have passed the canons of scientific acceptability. The emergence of a dominant paradigm signals scientific maturity and the acceptance of agreed-upon "standards" by which what has been referred to as "normal" scientific research can proceed. These "standards'' remain in force unless the paradigm is overturned. Revolutionary science is what 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 69 STRATEGIC PARTNERING, AND LICENSING DECISIONS overturns normal science, as when the Copernican theories of astronomy overturned those of Ptolemy in the seventeenth century. Abernathy and Utterback (1978), Dosi (1982), and Utterback (in this volume) provide treatment of the technological evolution of an industry in ways that parallel Kuhnian notions of scientific evolution (Kuhn, 1970). In the early stages of industrial development, product designs are fluid, manufacturing processes are loosely and adaptively organized, and generalized capital is used in production. Competition among firms manifests itself in competition among designs, which are markedly different from each other. This might be called the pre-paradigmatic stage of an industry. After considerable trial and error in the marketplace, one design or a narrow class of designs begins to emerge as the most promising. Such design must be able to meet a set of user needs in a relatively complete fashion. The Model T Ford, the IBM System/360, and the Douglas DC-3 are examples of dominant designs in the automobile, computer, and aircraft industries, respectively. Once a dominant design emerges, competition shifts to price and away from design. Competitive success then shifts to a new set of variables. Scale and learning become much more important, and specialized capital is deployed as incumbents seek to lower unit costs through exploiting economies of scale and learning. Reduced uncertainty over product design provides an opportunity to amortize specialized long-lived investments. Innovation is not necessarily halted once the dominant design emerges; as Clarke (1985) points out, it can occur at a lower level in the design hierarchy. For instance, a ''v" cylinder configuration emerged in automobile engine blocks during the 1930s with the Ford V-8 engine. Niches were quickly found for it. Moreover, once the product design stabilizes, there is likely to be a surge of process innovation as producers attempt to lower production costs for the new product (see Figure 2). The Abernathy-Utterback framework does not characterize all industries. It seems better suited to mass markets, in which consumer tastes are relatively homogeneous, than to small niche markets where the absence of scale and learning economies attaches a much lower penalty to multiple designs. For these niche markets, generalized equipment will be used in production. The emergence of a dominant design is a watershed that holds great significance for the distribution of profits between innovator and follower. The innovator may have been responsible for the fundamental scientific breakthroughs as well as the basic design of the new product. However, if imitation is relatively easy, imitators may enter the fray, modifying the product in important ways, yet relying on the fundamental designs pioneered by the innovator. When the game of musical chairs stops and a 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 70 STRATEGIC PARTNERING, AND LICENSING DECISIONS dominant design emerges, the innovator might well end up in a disadvantageous position relative to a follower. Hence, when imitation is coupled with design modification before the emergence of a dominant design, followers have a good chance that their modified product will be anointed as the industry standard, often to the great disadvantage of the innovator. Figure 2 Innovation over the product/industry life cycle. Complementary Assets Let the unit of analysis be an innovation. An innovation consists of technical knowledge about how to do something better than the existing state of the art. Assume that the know-how in question is partly codified and partly tacit. For such know-how to generate profits, it must be sold or used in the market. In almost all cases, the successful commercialization of an innovation requires that the know-how in question be used in conjunction with other capabilities or assets. Services such as marketing, competitive manufacturing, and after-sales support are almost always needed. These services are often obtained from complementary assets that are specialized. For example, the commercialization of a new drug is likely to require the dissemination of information over a specialized information channel. In some cases, as when the innovation is systemic, the complementary. assets may be other parts of a system. For instance, computer hardware typically 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 71 STRATEGIC PARTNERING, AND LICENSING DECISIONS requires specialized software, both for the operating system and for applications. Even when an innovation is autonomous, as with plug-compatible components, certain complementary capabilities or assets will be needed for successful commercialization. Figure 3 summarizes this relationship schematically. An important distinction is whether the assets required for least-cost production and distribution are specialized to the innovation. Figure 4 illustrates differences between complementary assets that are genetic, specialized, and cospecialized. Genetic assets are general-purpose assets that need not be tailored to the innovation in question. Specialized assets are those where there is unilateral dependence between the innovation and the complementary asset. Cospecialized assets are those for which there is a bilateral dependence. For instance, specialized repair facilities were needed to support the introduction of the rotary engine by Mazda. These assets are cospecialized because of the mutual dependence of the innovation on the repair facility. Containerization similarly required the deployment of some cospecialized assets in ocean shipping and terminals. However, the dependence of truck Figure 3 Complementary assets needed to commercialize an innovation. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 72 STRATEGIC PARTNERING, AND LICENSING DECISIONS ing on containerized shipping was less than that of containerized shipping on trucking, as trucks can convert from containers to flatbeds at low cost. An example of a generic asset is the manufacturing facilities needed to make running shoes. Generalized equipment can be employed in the main, exceptions being the molds for the soles. Figure 4 Complementary assets: generic, specialized, and cospecialized. IMPLICATIONS FOR PROFITABILITY These three concepts can now be related in a way that sheds light on the imitation process and the distribution of profits between innovator and follower. We begin by examining tight appropriability regimes. Tight Appropriability Regimes In those few instances when the innovator has an ironclad patent or copyright protection, or when the nature of the product is such that trade 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 73 STRATEGIC PARTNERING, AND LICENSING DECISIONS secrets effectively deny imitators access to the relevant knowledge, the innovator is almost assured that the innovation can be translated into market value for some period of time. Even if the innovator does not possess the desirable endowment of complementary costs, ironclad protection of intellectual property gives the innovator the time to acquire these assets. If these assets are genetic, a contractual relationship may well suffice, and the innovator may simply license the technology. Specialized R&D firms are viable in such an environment. Universal Oil Products, an R&D firm that developed refining processes for the petroleum industry, is a case in point. If, however, the complementary assets are specialized or cospecialized, contractual relationships are exposed to hazards, because one or both parties will have to commit capital to certain irreversible investments, which will be valueless if the relationship between innovator and licensee breaks down. Accordingly, the innovator may find it prudent to expand by acquiring or developing specialized and cospecialized assets. Fortunately, the factors that render imitation difficult will enable the innovator to build or acquire those complementary assets without competing with imitators for their control. Competition from imitators is muted in this type of regime, which sometimes characterizes the petrochemical industry. In this industry, the protection offered by patents is fairly easily enforced. One factor assisting the licensee in this regard is that most petrochemical processes are designed around a variety of catalysts that can be kept proprietary. An agreement not to analyze the catalyst can be extracted from licensees, affording extra protection. However, even if such requirements are violated by licensees, the innovator is still well positioned, as the most important properties of a catalyst are related to its physical structure, and the process for generating this structure cannot be deduced from structural analysis alone. Every chemical-reaction technology a company acquires is thus accompanied by an ongoing dependence on the innovating company for the catalyst appropriate to the plant design. Failure to comply with the licensing contract can thus result in a cutoff in the supply of the catalyst and possibly closure of the facility. Similarly, if the innovator comes to market in the pre-paradigmatic phase with a sound product concept but the wrong design, a tight appropriability regime will afford the innovator the time needed to perform the trials needed to get the design right. As discussed earlier, the best initial design concepts often turn out to be hopelessly wrong, but if the innovator is protected by an impenetrable thicket of patents, or has technology that is difficult to copy, then the market may well afford the innovator the necessary time to develop the right design before being eclipsed by imitators. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 74 STRATEGIC PARTNERING, AND LICENSING DECISIONS Loose Appropriability Tight appropriability is the exception rather than the rule. Therefore, innovators must turn to business strategy if they are to keep imitators at bay. The nature of the competitive process will depend on whether the industry is in the paradigmatic or pre-paradigmatic phase. Pre-Paradigmatic Phase In the pre-paradigmatic phase, the innovator must be careful to let the basic design "float" until there is sufficient evidence that the design is likely to become the industry standard. In some industries there may be little opportunity for product modification. In microelectronics, for example, designs become locked in when the circuitry is chosen. Product modification is limited to debugging and modifying software. An innovator must begin the design process anew if the product does not fit the market wen. In some respects, however, the selection of designs is dictated by the need to meet compatibility standards so that new hardware can be used with existing applications software. In one sense, therefore, the design issue for the microprocessor industry today is relatively straightforward: deliver greater power and speed while meeting the computer industry standards of the existing software base. However, from time to time windows of opportunity emerge for the introduction of entirely new families of microprocessors that will define a new industry and software standard. In these instances, basic design parameters are less well defined and can be permitted to "float" until market acceptance is apparent. The early history of the automobile industry exemplifies the importance of selecting the fight design in the pre-paradigm tic stages. None of the early producers of steam-powered cars survived the early shakeout when the internal combustion engine in a closed-body automobile emerged as the dominant design. The steamer, nevertheless, had numerous early virtues, such as reliability, which the cars with internal combustion engines could not deliver. The British fiasco with the Comet I is also instructive. DeHavilland had picked an early design that had both technical and commercial flaws. By moving into production, significant irreversibilities and loss of reputation hobbled de Havilland to such a degree that it was unable to convert to the Boeing design that subsequently emerged as dominant. It was not even able to occupy second place, which went instead to Douglas. As a general principle, it appears that innovators in loose appropriability regimes need to be intimately coupled to the market so that user needs can affect designs. When multiple parallel and sequential prototyping is 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 75 STRATEGIC PARTNERING, AND LICENSING DECISIONS feasible, it has clear advantages. Generally such an approach is prohibitively costly. When development costs for a large commercial aircraft exceed a billion dollars, variations on a theme are all that is possible. Hence, assessing the probability that an innovator will enter the paradigmatic phase possessing the dominant design is problematic. The probabilities increase, the lower the relative cost of prototyping and the more tightly coupled the firm is to the market. The latter is a function of organizational design and can be influenced by managerial choices. The former is embedded in the technology and cannot be influenced, except in minor ways, by managerial decisions. Consequently, in industries with large costs for development and prototyping—hence significant irreversibilities—and where innovation of the product concept is easy, the probability that the innovator will emerge as a winner at the end of the pre-paradigmatic stage is low. Paradigmatic Stage In the pre-paradigmatic phase, complementary assets do not loom large. Rivalry focuses on trying to identify the design that will be dominant. Production volumes are low, and there is little to be gained in deploying specialized assets, as scale economies are unavailable and price is not a principal competitive factor. As the leading design or designs begin to be revealed by the market, however, volumes increase and opportunities for economies of scale induce firms to begin gearing up for mass production by acquiring specialized tooling and equipment and possibly specialized distribution as well. Since these investments impose significant irreversibilities, producers are likely to proceed with caution. Islands of specialized capital will begin to appear in an industry that otherwise features a sea of general-purpose manufacturing equipment. But as the terms of competition begin to change, and prices become increasingly important, access to complementary assets becomes critical. Since the core technology is easy to imitate, by assumption, commercial success swings upon the terms and conditions affecting access to the required complementary assets. It is at this point that specialized and cospecialized assets become critically important. Generalized equipment and skills, almost by definition, are always available in an industry, and even if they are unavailable, they do not entail significant irreversibilities. Accordingly, firms have easy access to this type of capital, and, even if the relevant assets are not available in sufficient quantity, they can easily be put in place as this involves few risks. Specialized assets, on the other hand, imply significant irreversibilities and cannot be easily acquired by contract, as the risks are 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 76 STRATEGIC PARTNERING, AND LICENSING DECISIONS significant for the party making the dedicated investment. The firms that control the cospecialized assets, such as distribution channels and specialized manufacturing capacity, are clearly in an advantageous position relative to an innovator. Indeed, in the rare instances in which incumbent firms possess an airtight monopoly over specialized assets, and the innovator is in a regime of loose appropriability, all of the profits from the innovation could conceivably accrue to those firms, who should be able to get the upper hand. Even when the innovator does not face competitors or potential competitors who control key assets, the innovator may still be disadvantaged. For instance, the technology embedded in cardiac pacemakers was easy to imitate, so competitive outcomes quickly came to be determined by who had easiest access to the complementary assets—in this case, specialized marketing. A similar situation has recently arisen in the United States with respect to personal computers. As an industry participant recently observed: There are a huge number of computer manufacturers, companies that make peripherals (e.g., printers, hard disk drives, floppy disk drives), and software companies. They are all trying to get marketing distributors became they cannot afford to call on all of the U.S. companies directly. They need to go through retail distribution channels, such as Businessland, in order to reach the marketplace. The problem today, however, is that many of these companies are not able to get shelf space and thus are having a very difficult time marketing their products. The point of distribution is where the profit and the power are in the marketplace today. [Norman, 1986, p. 438] CHANNEL STRATEGY ISSUES The preceding analysis indicates how access to complementary assets, such as manufacturing and distribution, on competitive terms is critical if the innovator is to avoid handing over most of the profits to imitators, or to the owners of the complementary assets that are specialized or cospecialized to the innovation. It is now necessary to delve deeper into the control structure that the innovator ideally will establish over these critical assets. There are many possible channels that could be employed. At one extreme the innovator could integrate into all of the necessary complementary assets, but complete integration is likely to be unnecessary and also prohibitively expensive. It is important to recognize that the variety of assets and competences needed is likely to be quite large, even for only modestly complex technologies. To produce a personal computer, for instance, a company needs access to expertise in semiconductor, display, disk drive, networking, and keyboard technologies, among others. No 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 77 STRATEGIC PARTNERING, AND LICENSING DECISIONS company can keep pace in all of these areas by itself. At the other extreme, the innovator could try to gain access to these assets through straightforward contractual relationships (for example, component supply contracts, fabrication contracts, service contracts). In many instances such contracts may suffice, although they sometimes expose the innovator to hazards and dependencies that might otherwise be avoided. Between the fully integrated and full contractual extremes are many intermediate forms and channels. An analysis of the properties of the two extreme forms is presented below. A brief synopsis of mixed modes then follows. Contractual Modes The advantages of a contractual solution—whereby the innovator signs a contract, such as a license, with independent suppliers, manufacturers, or distributors—are obvious. The innovator will not have to make the up-front capital expenditures needed to build or buy the assets in question. This reduces risks as well as cash requirements. Contracting rather than integrating is likely to be the optimal strategy when the innovator's appropriability regime is tight and the complementary assets are available in competitive supply (that is, there is adequate capacity and a choice of sources). Both conditions apply in the petrochemical industry, for instance, so an innovator does not need to be integrated to be successful. Consider, first, the appropriability regime. As discussed earlier, the protection offered by patents is fairly easily enforced, particularly for process technology, in the petrochemical industry. Given the advantageous feedstock prices available to hydrocarbon- rich petrochemical exporters, and the appropriability regime characteristic of this industry, there is neither incentive nor advantage in owning the complementary assets (production facilities), as they are not typically highly specialized to the innovation. Union Carbide appears to realize this and has recently adjusted its strategy accordingly. Essentially, Carbide is placing its existing technology into a new subsidiary, Engineering and Hydrocarbons Service. The company is engaging in licensing and offers engineering, construction, and management services to customers who want to take their feedstocks and integrate them forward into petrochemicals. But Carbide itself appears to be backing away from an integration strategy. Chemical and petrochemical product innovations are not as easily protected as process technology is, which should raise new challenges to innovating firms in developed nations as they attempt to shift out of commodity petrochemicals. There are already numerous examples of new products that made it to the marketplace, filled a customer need, but never 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 78 STRATEGIC PARTNERING, AND LICENSING DECISIONS generated competitive returns to the innovator because of imitation. For example, in the 1960s the Dow Chemical Company decided to start manufacturing rigid polyurethene foam. It was quickly imitated, however, by many small firms that had lower costs.2 The absence of low-cost manufacturing capability left Dow vulnerable. Contractual relationships can bring added credibility to the innovator, especially if the innovator is relatively unknown when the contractual partner is established and viable. Indeed, arms-length contracting that embodies more than a simple buy-sell agreement is becoming so common, and is so multifaceted, that the term strategic partnering has been devised to describe it. Even large companies such as IBM are now engaging in it. For IBM, partnering buys access to new technologies, enabling a company to ''learn things we couldn't have learned without many years of trial and error."3 IBM's arrangement with Microsoft Corporation for the use of MS-DOS operating system software on the IBM PC facilitated the timely introduction of IBM's personal computer into the market. Smaller, less integrated companies are often eager to sign on with established companies because of the name recognition and reputation spillovers. For instance, Cipher Data Products, Inc., contracted with IBM to develop a low-priced version of IBM's 3480 half-inch streaming cartridge drive, which is likely to become the industry standard. As Cipher management points out, "one of the biggest advantages to dealing with IBM is that, once you've created a product that meets the high quality standards necessary to sell into the IBM world, you can sell into any arena."4 Similarly, IBM's contract with Microsoft "meant instant credibility" to Microsoft (McKenna, 1985, p. 94). It is most important to recognize, however, that strategic (contractual) partnering, which is currently fashionable, holds certain hazards, particularly for the innovator, when the innovator is trying to use contracts to acquire specialized capabilities. First, it may be difficult to induce suppliers to make costly irreversible commitments that depend for their success on the success of the innovation. To expect suppliers, manufacturers, and distributors to do so is to invite them to take risks along with the innovator. The problem this poses for the innovator is similar to the problems associated with attracting venture capital. The innovator must persuade its prospective partner that the risk is a good one. The situation is open to opportunistic abuses on both sides. The innovator has incentives to overstate the value of the innovation, while the supplier has incentives to "run with the technology" should the innovation be a success. Instances of irreversible capital commitments by both parties nevertheless exist. Apple's Laserwriter—a laser printer that allows PC users to produce near- typeset-quality text and art department graphics—is a case 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 79 STRATEGIC PARTNERING, AND LICENSING DECISIONS in point. Apple persuaded Canon, Inc., to participate in the development of the Laserwriter by providing subsystems from its copiers—but only after Apple contracted to pay for a certain number of copier engines and cases. In short, Apple accepted a good deal of the financial risk to induce Canon to assist in the development and production of the Laserwriter. The arrangement appears to have been prudent, yet there were clearly hazards for both sides. It is difficult to write, execute, and enforce complex development contracts, particularly when the design of the new product is still ''floating." Apple was exposed to the risk that its coinnovator Canon would fail to deliver, and Canon was exposed to the risk that the Apple design and marketing effort would not succeed. Still, Apple's alternatives may have been limited, inasmuch as it did not command the requisite technology to "go it alone." In short, the current euphoria over strategic partnering may be partially misplaced. The advantages are being stressed (for example, McKenna, 1985) without a balanced presentation of costs and risks. Briefly, there is the risk that the partner will not perform according to the innovator's perception of what the contract requires; there is the added danger that the partner may imitate the innovator's technology and attempt to compete with the innovator. This latter possibility is particularly acute if the provider of the complementary asset is uniquely situated with respect to the complementary asset in question and has the capacity to imitate the technology, which the innovator is unable to protect. The innovator will then find that it has created a competitor who is better positioned than the innovator to take advantage of the market opportunity at hand. Business Week has expressed concerns along these lines in its discussion of the "hollow corporation."5 It is important to bear in mind, however, that contractual or partnering strategies in certain cases are ideal. If the innovator's technology is well protected, and if what the partner has to provide is a "generic" capacity available from many potential partners, then the innovation will be able to maintain the upper hand while avoiding the costs of duplicating downstream capacity. Even if the partner fails to perform, adequate alternatives exist (by assumption, the partner's capacities are commonly available) so the innovator's efforts to successfully commercialize the technology ought to proceed profitably. Integration Modes Integration, which by definition involves ownership, is distinguished from pure contractual modes in that it typically facilitates incentive alignment and tighter organizational control (Williamson, 1985). An innovator who owns rather than rents the complementary assets needed to com 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 80 STRATEGIC PARTNERING, AND LICENSING DECISIONS mercialize is in a position to capture spillover benefits stemming from increased demand for the complementary assets caused by the innovation. Indeed, an innovator might be in the position, at least before the innovation is announced, to buy up capacity in the complementary assets, possibly to great subsequent advantage. If futures markets exist, though generally speaking they do not, taking forward positions in the complementary assets may suffice to capture much of the spillover. Even after the innovation is announced, the innovator might still be able to build or buy complementary capacities at competitive prices if the innovation has ironclad legal protection (that is, if the innovation is in a tight appropriability regime). However, if the innovation is not tightly protected and once out is easy to imitate, then securing control of complementary capacities is likely to be the key success factor, particularly if those capacities are in fixed supply—so-called bottlenecks. Distribution and specialized manufacturing competences often become bottlenecks. As a practical matter, however, an innovator may not have the time to acquire or build the complementary assets that ideally would be desirable. This is particularly true when imitation is easy, so that timing becomes critical. Additionally, the innovator may not have the financial resources to proceed. The implications of timing and cash constraints are summarized in Figure 5. Accordingly, in loose appropriability regimes innovators need to rank complementary assets according to their importance. If the complementary assets are critical, ownership is warranted, although if the firm is cash constrained a minority position may well be a sensible approach. When imitation is easy, strategic moves to build or buy specialized complementary assets must occur with due reference to the moves of competitors. There is no point in attempting to build a specialized asset, for instance, if one's imitators can do it faster and cheaper. It should be self-evident that if the innovator is already a large enterprise with control over many of the relevant complementary assets, integration is not likely to be the issue it might otherwise be, as the innovating firm will already control many of the relevant specialized and cospecialized assets. However, in industries experiencing rapid technological change, it is unusual that a single company has the full range of expertise needed to bring advanced products to market in a timely and cost-effective way. Hence, the integration issue is of concern to both large and small firms. Integration Versus Contract Strategies: An Analytic Summary Figure 6 summarizes some of the relevant considerations in the form of a decision flow chart. It indicates that a profit-seeking innovator faced 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 81 STRATEGIC PARTNERING, AND LICENSING DECISIONS Figure 5 Specialized complementary assets and loose appropriability: integration calculus. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 82 STRATEGIC PARTNERING, AND LICENSING DECISIONS with weak protection of intellectual property and the need to access specialized complementary assets or capabilities is forced to expand through integration to prevail over imitators. Put differently, innovators who develop new products that possess poor protection of intellectual property but require specialized complementary capacities are more likely to parlay their technology into a commercial advantage rather than see it prevail in the hands of imitators. Figure 6 makes it apparent that difficult strategic decisions arise when the appropriability regime is loose and when specialized assets are critical to profitable commercialization. These situations are common and require that a thorough assessment of competitors be part of the innovator's strategic assessment of opportunities and threats. Figure 7 carries this discussion a step further and considers only situations where commercialization requires certain specialized capabilities. It shows the appropriate strategies for the innovators and predicts the expected outcomes for the various players. Three classes of players are of interest: innovators, imitators, and the owners of cospecialized assets (for example, distributors). All three can potentially benefit or lose from the innovation process. The latter can potentially benefit from the additional business that the innovation may direct in the asset owner's direction. Should the asset turn out to be a bottleneck with respect to commercializing the innovation, the owner of the bottleneck facilities is obviously in a position to extract profits from the innovator or the imitators. The vertical axis in Figure 7 measures how those who possess the technology (the innovator or possibly the imitators) are positioned with respect to those firms that possess required specialized assets. The horizontal axis measures the "tightness" of the appropriability regime, tight regimes being evidenced by ironclad legal protection coupled with technology that is difficult to copy; loose regimes offer little in the way of legal protection, and the essence of the technology, once released, is transparent to the imitator. Loose regimes are further subdivided according to how the innovator and imitators are positioned in relation to each other. This is likely to be a function of factors such as lead time and prior positioning in the requisite complementary assets. Figure 7 makes it apparent that even when firms pursue the optimal strategy, other industry participants may take the jackpot. This possibility is unlikely when the intellectual property in question is tightly protected. The only serious threat to the innovator is where a specialized complementary asset is "locked up," a possibility recognized in cell 4. This can rarely be done without the cooperation of government. But it frequently occurs, as when a foreign government closes access to a foreign market, 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 83 STRATEGIC PARTNERING, AND LICENSING DECISIONS Figure 6 Flow chart for integration versus contract design. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 84 STRATEGIC PARTNERING, AND LICENSING DECISIONS forcing the innovators to license to foreign firms, but with the government effectively cartelizing the potential licensees. With weak intellectual property protection, however, it is clear that the innovator will often lose out to imitators or asset holders, even when the innovator is pursuing the appropriate strategy (cell 6). Clearly, incorrect strategies can compound problems. For instance, if innovators integrate when they should contract, Figure 7 Optimal contract and integration strategies and outcomes for innovators: specialized asset case. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 85 STRATEGIC PARTNERING, AND LICENSING DECISIONS a heavy commitment of resources will be incurred for little if any strategic benefit, thereby exposing the innovator to even greater losses than would otherwise be the case. On the other hand, if an innovator tries to contract for the supply of a critical capability when it should build the capability itself, it may well find it has nurtured an imitator better able to serve the market. Mixed Modes The real world rarely provides extreme or pure cases. Decisions to integrate or license involve trade-offs, compromises, and mixed approaches. It is not surprising, therefore, that the real world is characterized by mixed modes of organization, involving judicious blends of integration and contracting. Sometimes mixed modes represent transitional phases. For instance, because of the convergence of computer and telecommunication technology, firms in each industry are discovering that they often lack the technical capabilities needed in the other. Since the technological interdependence of the two requires collaboration among those who design different parts of the system, intense cross-boundary coordination and information flows are needed. For separate enterprises, agreement must be reached on complex protocol issues among parties who see their interests differently. Contractual difficulties can be anticipated, as the selection of common technical protocols among the parties will often be followed by transaction-specific investments in hardware and software. There is little doubt that this was the motivation behind IBM's 1983 purchase of 15 percent of the Rolm Corporation, manufacturer of business communications systems. This position was expanded to 100 percent in 1984. IBM's stake in Intel Corporation, which began with a 12 percent purchase in 1982, is most probably not a transitional phase leading to 100 percent purchase, because both companies realized that the two corporate cultures are not compatible, and IBM may not be as impressed with Intel's technology as it once was. The Cat Scanner And The IBM PC: Insights From The Framework EMI's failure to reap significant returns from the CAT scanner can be explained in large measure by reference to the concepts developed above. The scanner that EMI developed was of a technical sophistication much higher than would normally be found in a hospital, requiring a high level of training support and servicing. EMI did not possess these capabilities, could not easily contract for them, and was slow to realize their importance. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 86 STRATEGIC PARTNERING, AND LICENSING DECISIONS It most probably could have formed a partnership with a company like Siemens to gain access to the requisite capabilities. Its failure to do so was a strategic error compounded by the limited protection afforded by the law for the intellectual property embodied in the scanner. Although subsequent court decisions have upheld some of EMI's patent claims, once the product was in the market it could be reverse engineered and its essential features copied. Two competitors, General Electric and Technicare, already possessed the complementary capabilities that the scanner required, and they were also technologically capable. In addition, both were experienced marketers of medical equipment and had reputations for quality, reliability, and service. GE and Technicare were thus able to commit their R&D resources to developing a competitive scanner and improving on the EMI scanner where they could while they rushed to market. GE began taking orders in 1976 and soon after made inroads on EMI's lead. In 1977 concern for rising health care costs caused the Carter administration to introduce "certificate of need" regulation, which required approval by the Department of Health, Education, and Welfare for expenditures on big ticket items like CAT scanners. This severely cut the size of the available market. By 1978 EMI had lost the leadership in market share to Technicare, who was in turn quickly overtaken by GE. In October 1979 Geoffrey Hounsfield of EMI shared the Nobel Prize for invention of the CAT scanner. Despite this honor, and the public recognition of EMI's role in bringing this medical breakthrough to the world, the collapse of its scanner business forced EMI in the same year into the arms of a rescuer, Thorn Electrical Industries, Ltd. GE subsequently acquired what was EMI's scanner business from Thom.6 Though royalties continued to flow to EMI, the company had failed to capture the largest pan of the profits generated by the innovation it had pioneered and successfully commercialized. If EMI illustrates how a company with outstanding technology and an excellent product can fail to profit from innovation while the imitators succeed, the story of the IBM PC indicates how a new product representing only a modest technological advance can yield remarkable returns to the developer. The IBM PC, introduced in 1981, succeeded despite the fact that its architecture was ordinary and its components standard. Philip Estridge's design team in Boca Raton, Florida, decided to use existing technology rather than the state of the art to produce a solid, reliable microcomputer. With a 1-year mandate to develop a PC, Estridge's team could do little else. 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 87 STRATEGIC PARTNERING, AND LICENSING DECISIONS However, the IBM PC did use what at the time was a new 16-bit microprocessor (the Intel 8088) and a new disk operating system (DOS) adapted for IBM by Microsoft. Other than the microprocessor and the operating system, the IBM PC incorporated existing microcomputer "standards" and used off-the-shelf parts from outside vendors. IBM did write its own basic input/ output system (BIOS), which is embedded in a read-only memory chip, but this was a relatively straightforward programming exercise. The key to the PC's success was not the technology. It was the set of complementary assets that IBM either had or quickly assembled around the PC. To expand the market for PCs, there was a clear need for an expandable, flexible microcomputer system with extensive applications of software. IBM could have based its PC system on its own patented hardware and copyrighted software. Such an approach would cause complementary products to be cospecialized, forcing IBM to develop peripherals and a comprehensive library of software in a short time. Instead, IBM adopted what might be called an induced contractual approach. By adopting an open system architecture, as Apple had done, and by making technical information about the operating system publicly available, IBM induced a spectacular output of software by third-party suppliers. IBM estimated that by mid-1983, at least 3,000 hardware and software products were available for the PC.7 Put differently, IBM pulled together the complementary assets, particularly software, required for success and did not even use contracts, let alone integration. This was despite the fact that the software developers were creating assets that were in part cospecialized to the IBM PC, at least in the first instance. Several special considerations made this approach a reasonable risk for the software writers. A critical element was IBM's name and corn-raiment to the project. The reputation behind the letters I, B, M is perhaps the greatest cospecialized asset the company possesses. The name implied that the product would be marketed and serviced in the IBM tradition. It guaranteed that PC- DOS would become an industry standard, so that the software business would not be solely dependent on IBM, because emulators were sure to enter. It guaranteed access to retail distribution outlets on competitive terms. The consequence was that IBM was able to take a product that was at best a modest technological accomplishment and turn it into a fabulous commercial success. The case demonstrates the role of complementary assets in determining outcomes. Though the success of the IBM PC is ongoing, the appearance of machines compatible with the IBM PC (IBM compatibles and "clones") has 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 88 STRATEGIC PARTNERING, AND LICENSING DECISIONS somewhat attenuated PC market growth for IBM. The emergence and rapid acceptance of the IBM PC established a market-based software standard. Given IBM's reputation and the quality of the product, the emergence of a market standard was predictable, as was increased price competition as competitors focused on cost reduction and performance enhancement. The fact that IBM is no longer the overwhelmingly dominant PC manufacturer—possibly because of its price umbrella and modest rate of performance improvement—does not diminish the lesson of the IBM PC program with regard to capturing returns from innovation. Despite competition from compatibles and clones, IBM's return on investment must surely have been attractive. IMPLICATIONS FOR R&D STRATEGY, INDUSTRY STRUCTURE, AND TRADE POLICY Allocating R&D Resources The analysis so far assumes that the firm has developed an innovation for which a market exists. It indicates the strategies that the firm must follow to maximize its share of industry profits relative to imitators and other competitors. There is no guarantee of success even if optimal strategies are followed. The innovating firm can improve its total return to R&D, however, by adjusting its R&D investment portfolio to maximize the probability that the technological discoveries that emerge will be easy to protect with existing property law or will require for commercialization cospecialized assets already within the firm's repertoire of capabilities. Put differently, if an innovating firm does not target its R&D resources toward new products and processes that it can commercialize advantageously relative to potential imitators or followers, then it is unlikely to profit from its investment in R&D. In this sense, a firm's history —and the assets it already has in place—ought to condition its R&D investment decisions. Clearly, an innovating firm with considerable assets already in place is free to strike out in new directions, so long as it is aware of the kinds of capabilities required to commercialize the innovation. It is therefore clear that the R&D investment decision cannot be divorced from the strategic analysis of markets and industries, and the firm's position in them. Small Firm Versus Large Firm Comparisons Business commentators frequently remark that many small entrepreneurial firms that generate new, commercially valuable technology fail at 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 89 STRATEGIC PARTNERING, AND LICENSING DECISIONS the same time that a large multinational firm, even with a less meritorious record in innovation, will survive and prosper. One explanation of this phenomenon is now clear. Large firms are more likely to possess the relevant specialized and cospecialized assets at the time a new product is introduced. They can therefore do a better job of using their technology, however meager, to maximum advantage. Small domestic firms are less likely to have the relevant specialized and cospecialized assets within their boundaries. They must therefore incur the expense of trying either to build the necessary assets or to develop coalitions with competitors or with owners of the assets. Regimes Of Appropriability And Industry Structure In industries where legal methods of protection are effective, or where new products are just hard to copy, the strategic necessity for innovating firms to obtain cospecialized assets would appear to be less compelling than in industries where legal protection is weak. In cases where legal protection is weak or nonexistent, the control of cospecialized assets will be needed for long- run survival. In this regard, it is instructive to examine the U.S. drug industry (Temin, 1979). In the 1940s, the U.S. Patent Office began to grant patents on certain natural substances that involved difficult extraction procedures. Thus, in 1948 Merck received a patent on streptomycin, which is a natural substance. However, it was not the extraction process but the drug itself that received the patent. Hence, patents were important to the drug industry, but they did not prevent imitation as, in some cases, just changing one molecule would enable a company to come up with a similar substance not violating the patent (Temin, 1979, p. 436). Had patents been more inclusive—and this is not to suggest that they should be—licensing would have been an effective mechanism for Merck to profit from its innovation. The emergence of close substitutes for patented drags, coupled with FDA regulation that had the effect of reducing the elasticity of demand for drugs, placed high rewards on a strategy of product differentiation. This strategy required extensive marketing, including a sales force that could directly contact doctors, who were the purchasers of drugs through their ability to create prescriptions.8 The result was exclusive production (that is, the earlier industry practice of licensing was dropped) and forward integration into marketing (the relevant cospecialized asset). Generally, if legal protection of the innovators' profits is secure, innovating firms can select their boundaries according to their ability to identify user needs and respond to those needs through research and development. The weaker the legal methods of protection, the greater the 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 90 STRATEGIC PARTNERING, AND LICENSING DECISIONS incentive to obtain the relevant cospecialized assets. Hence, as industries in which legal protection is weak begin to mature, integration into innovation- specific cospecialized assets will occur. Often this will take the form of backward, forward, and lateral integration. (Conglomerate integration is not part of this phenomenon.) For example, IBM's purchase of Rolm can be seen as a response to the impact of technological change on the identity of the cospecialized assets relevant to IBM's future growth. Industry Maturity, New Entry, And History As technologically progressive industries mature, and a greater proportion of the relevant cospecialized assets are brought under the corporate umbrellas of incumbents, new entry becomes more difficult. Moreover, when it does occur it is more likely to include the early formation of coalitions. Incumbents will own the cospecialized assets, and new entrants will find it necessary to forge links with them. Here lies the explanation for the sudden surge in strategic partnering now occurring internationally, and particularly in the computer and telecommunications industry. Note that this change should not be interpreted in anticompetitive terms. Given existing industry structure, coalitions ought to be seen not as attempts to stifle competition, but as mechanisms for lowering entry requirements for innovators. In industries in which a technological change has occurred and required deployment of specialized or cospecialized assets, a configuration of firm boundaries that no longer have compelling efficiencies may well have arisen. Considerations that once dictated integration may no longer hold, yet there may not be strong forces leading to divestiture. Hence existing firm boundaries in some industries—especially those where the technological trajectory and attendant specialized asset requirements have changed—may be fragile. In short, history is important in understanding the structure of the modern business enterprise. Existing firm boundaries cannot always be assumed to have an obvious rationale in relation to today's requirements. The Importance Of Manufacturing To International Competitiveness Practically all forms of technological know-how must be embedded in goods and services to yield value to the consumer. An important policy issue for the innovating nation is whether the identity of the firms and nations performing this function is important. In a world of tight appropriability and zero transactions cost—the world of neoclassical theory—it is a matter of indifference whether an innovating firm has an in-house manufacturing capability, domestic or foreign. The 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 91 STRATEGIC PARTNERING, AND LICENSING DECISIONS firm can engage in' arms-length contracting (patent licensing, know-how licensing, coproduction, and so on) for the sale of the output of the activity in which it has a comparative advantage (in this case R&D) and will maximize returns by specializing in what it does best. However, in a regime of loose appropriability, and especially where the requisite manufacturing assets are specialized to the innovation. which is often the case, participation in manufacturing may be necessary if an innovator is to appropriate the rents from its innovation. Hence, if an innovator's manufacturing costs are higher than those of an imitator, the innovator may well be put in the position of ceding the largest share of profits to the imitator. In a loose appropriability regime, low-cost imitator-manufacturers could capture all of the profits from innovation. In a loose appropriability regime where specialized manufacturing capabilities are necessary to produce new products, an innovator with a manufacturing disadvantage may find that an early advantage at the research and development stage has no commercial value. This is a potentially crippling situation unless the innovator is assisted by governmental processes. For example, one reason why U.S. manufacturers did not capture the greatest part of the profits from the development of color TV, for which RCA was primarily responsible, is that RCA and its U.S. licensees were not competitive at manufacturing. In this context, concern that the decline of manufacturing threatens the entire economy appears to be well founded. A related implication is that as the technology gap closes, the basis of competition in an industry will shift to the cospecialized assets. This appears to be what is happening in microprocessors. Intel is no longer out ahead technologically. As Gordon Moore, CEO of Intel points out, ''Take the top 10 [semiconductor] companies in the world... and it is hard to tell at any time who is ahead of whom .... It is clear that we have to be pretty damn close to the Japanese from a manufacturing standpoint to compete.''9 It is not just that strength in one area is necessary to compensate for weakness in another. As technology becomes more public and less proprietary through easier imitation, strength in manufacturing and other areas is necessary to benefit from whatever technological advantages an innovator may possess. Put differently, the notion that the United States can adopt a "designer role" in international commerce while letting independent firms in countries such as Japan, Korea, Taiwan, or Mexico do the manufacturing is unlikely to be a successful strategy for the long run. This is because profits will accrue primarily to the low-cost manufacturers (by providing a larger sales base over which they can exploit their special skills). Where imitation is easy, and even where it is not, it is difficult to do business in the market 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 92 STRATEGIC PARTNERING, AND LICENSING DECISIONS for know-how (Teece, 1981). In particular, there are difficulties in pricing an intangible asset whose true performance features are difficult to predict. The trend in international business toward what Miles and Snow (1986) call dynamic networks—characterized by vertical disintegration and contracting- therefore ought to be viewed with concern. Dynamic networks, or hollow corporations, may reflect innovative organizational forms not so much as the disassembly of the modem corporation because of deterioration in manufacturing and other activities that complement technological innovation. Dynamic networks-may therefore signal not so much the rejuvenation of American enterprise as its piecemeal demise. How Trade And Investment Barriers Affect Innovators' Profits In regimes of loose appropriability, governments can move to shift the distribution of the gains from innovation away from foreign innovators and toward domestic firms by denying innovators ownership of specialized assets. The foreign firm, by assumption an innovator, will be left with the option of selling its intangible assets on the market for know-how if both trade and investment are foreclosed by government policy. This option may appear better than the alternative (no remuneration at all from the market in question). Licensing may then appear profitable, but only because access to the complementary assets is blocked by government. Thus, when an innovating firm generating profits needs access to complementary assets abroad, host governments, by limiting access, can sometimes milk the innovators for a share of the profits, particularly that portion that originates from sales in the host country. However, the ability of host governments to do so depends on the importance of the host country's assets to the innovator. If the cost and infrastructure characteristics of the host country are such that it is the world's lowest cost manufacturing site, and if domestic industry is competitive, then by acting as a monopsonist the government of the host country ought to be able to adjust the terms of access to the complementary assets to appropriate a greater share of the profits generated by the innovation. 10 If, on the other hand, the host country offers no unique complementary assets except access to its own market, restrictive practices by the government will only redistribute profits with respect to domestic rather than worldwide sales. Implications For The International Distribution Of The Benefits From Innovation Thus, it is clear that innovators who do not have access to the relevant specialized and cospecialized assets may end up ceding profits to imitators 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 93 STRATEGIC PARTNERING, AND LICENSING DECISIONS and other competitors, or to the owners of the specialized or cospecialized assets. Even when the innovator possesses the specialized assets, they may be located abroad. Foreign factors of production are thus likely to benefit from research and development occurring across borders. There is little doubt, for instance, that the inability of many U.S. multinationals to sustain competitive manufacturing in the United States results in declining returns to U.S. labor. Stockholders and top management probably do as well if not better when a multinational gains access to cospecialized assets in the firm's foreign subsidiaries. However, if there is unemployment in the factors of production supporting these assets, then the foreign factors of production will benefit from innovation originating beyond national borders. This shows how important it is that innovating nations maintain competence and competitiveness in the assets— especially manufacturing—that complement technological innovation. It also shows how important it is that innovating nations enhance the protection afforded worldwide to intellectual property. It must be recognized, however, that there are inherent limits to the legal protection of intellectual property and that business and national strategies are therefore likely to be critical in determining how the gains from innovation are shared' worldwide. By making the correct strategic decision, innovating firms can move to protect the interests of stockholders. But to ensure that domestic rather than foreign cospecialized assets capture the largest share of the externalities spilling over to complementary assets, the supporting infrastructure for those complementary assets must not be allowed to decay. In short, if a nation has prowess at innovation, then in the absence of ironclad protection for intellectual property, it must maintain well-developed complementary assets if it is to capture the spillover benefits from innovation. CONCLUSION This chapter has attempted to synthesize from recent research in industrial organization and strategic management a framework within which to analyze the distribution of the profits from innovation. The framework indicates that the boundaries of the firm are an important strategic variable for innovating firms. The ownership of complementary assets, particularly when they are specialized or cospecialized, helps establish who wins and who loses from innovation. Imitators can often outperform innovators if they are better positioned with respect to critical complementary assets. Hence, public policy aimed at promoting innovation must focus not only on R&D but also on complementary assets as well as the underlying infrastructure. If government decides to stimulate innovation, it is im 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 94 STRATEGIC PARTNERING, AND LICENSING DECISIONS portant to eliminate barriers to the development of complementary assets that are specialized or cospecialized to innovation. To fail to do so will cause a large portion of the profits from innovation to flow to imitators and other competitors. If these firms lie beyond one's national borders, there are obvious implications for the international distribution of income. When applied to world markets, results similar to those obtained from the "new trade theory" are suggested by the framework. In particular, tariffs and other restrictions on trade can in some cases injure innovating firms while simultaneously benefiting protected firms when they are imitators. However, the propositions suggested by the framework vary according to appropriability regimes, suggesting that economywide conclusions will be elusive. The policy conclusions for commodity petrochemicals, for instance, are likely to differ from those for semiconductors. The approach also suggests that the product life cycle model of international trade win play itself out differently in different industries and markets, in part according to appropriability regimes and the nature of the assets needed to convert a technological success into a commercial one. Whatever its limitations, the approach establishes that it is not so much the structure of markets as the structure of firms, particularly the scope of their boundaries, coupled with national policies on the development of complementary assets, that determines the distribution of the profits among innovators and imitator- followers. ACKNOWLEDGMENTS I wish to thank Raphael Amit, Harvey Brooks, Chris Chapin, Therese Flaherty, Richard Gilbert, Bruce Guile, Heather Haveman, Mel Horwitch, David Hulbert, Carl Jacobson, Michael Porter, Gary Pisano, Richard Rumelt, Richard Nelson, Raymond Vernon, and Sidney Winter for helpful discussions relating to the subject matter of this paper. I gratefully acknowledge the financial support of the National Science Foundation under grant no. SRS-8410556 to the Center for Research in Management, University of California, Berkeley. Versions of this paper were presented at a National Academy of Engineering symposium titled "World Technologies and National Sovereignty," February 1986; at a conference on innovation at the University of Venice, March 1986; and at seminars at the Massachusetts Institute of Technology and Harvard, Yale, and Stanford universities. Helpful comments received at these conferences and seminars are gratefully acknowledged. NOTES 1. The EMI story is summarized in Michael Martin, Managing Technological Innovation and Entrepreneurship (Reston, Va.: Reston Publishing Company, 1984). 

CAPTURING VALUE FROM TECHNOLOGICAL INNOVATION: INTEGRATION, 95 STRATEGIC PARTNERING, AND LICENSING DECISIONS 2. Executive Vice President, Union Carbide, Robert D. Kennedy, quoted in Chemical Week. 3. Comment attributed to Peter Olson III, IBM's director of business development, as reported in "The Strategy Behind IBM's Strategic Alliances," Electronic Business, October I, 1985, p. 126. 4. Comment attributed to Norman Farquhar, Cipher's vice president for strategic development, as reported in Electronic Business, October I, 1985, p. 128. 5. See Business Week, March 3, 1986, pp. 57-59. Business Week uses the term "hollow corporation" to describe a firm that lacks in-house manufacturing capability. 6. See "GE Gobbles a Rival in CT Scanners," Business Week, May 19, 1980. 7. F. Gens and C. Christiansen, "Could 1,000,000 IBM PC Users Be Wrong," Byte, November 1983, p. 88. 8. In the period before FDA regulation, all drugs other than narcotics were available without prescriptions. Since the user could purchase drugs directly, sales were price sensitive. Once prescriptions were required, this price sensitivity collapsed; doctors not only do not have to pay for the drugs, but in most cases they are unaware of the prices of the drugs they are prescribing. 9. "Institutionalizing the Revolution," Forbes, June 16, 1986, p. 35. 10. If the host country market structure is monopolistic, private actors might be able to achieve the same benefit. Government can force collusion of domestic enterprises to their mutual benefit. REFERENCES Abernathy, W. J., and J. M. Utterback. 1978. Patterns of industrial innovation. Technology Review 80:7(June-July):40-47. Clarke, K. B. 1985. The interaction of design hierarchies and market concepts in technological evolution. Research Policy 14:235-251. Dosi, G. 1982. Technological paradigms and technological trajectories. Research Policy 1 I (3):147-162. Kuhn, T. 1970. The Structure of Scientific Revolutions, 2nd ed. Chicago: University of Chicago Press. Levin, R., A. Klevorick, N. Nelson, and S. Winter. 1984. Survey research on R&D appropriability and technological opportunity. Yale University. Unpublished manuscript. McKenna, R. 1985. Market positioning in high technology. California Management Review XXVII:3(Spring):82-108 Miles, R. E., and C. C. Snow. 1986. Network organizations: New concepts for new forms. California Management Review XXVIII:3(Spring):62-73. Norman, D. A. 1986. Impact of entrepreneurship and innovations on the distribution of personal computers. Pp.437-439 in R. Landau and N. Rosenberg, eds., The Positive Sum Strategy. Washington, D.C.: National Academy Press. Teece, D. J. 1981. The market for know how and the efficient international transfer of technology. Annals of the American Academy of Political and Social Science 458(November):81-96 Temin, P. 1979. Technology, regulation, and market structure in the modem pharmaceutical industry. The Bell Journal of Economics 10(2):429-446. Williamson, O. E. 1985. Economic Institutions of Capitalism. New York: The Free Press.