Innovation in Global Industries: U.S. Firms Competing in a New World (Collected Studies) (2008)

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.

6 Pharmaceuticals Iain M. Cockburn Boston University and National Bureau of Economic Research INTRODUCTION Pharmaceuticals is a highly globalized industry, dominated by multinational companies that engage in significant business activity in many countries and whose products are distributed and marketed worldwide. Historically, the indus- try has been dominated by vertically integrated firms performing almost all of the activities in the value chain, from basic research through to sales and marketing. But it is far from clear that these activities should necessarily be geographically co-located; these firms have generally operated globally, with many aspects of their operations spanning several countries. In recent decades the industry has undergone dramatic structural changes, with the rise of the biotechnology sector, substantial growth in demand driven by demographics and substitution away from other therapeutic modalities such as surgery, and increased competition from globally active generic manufacturers. These changes have led to some degree of dis-integration and geographic dispersion, but innovative activity is nonetheless highly geographically concentrated, reflecting the economic sig- nificance of factors such as localized knowledge spillovers and the strength of patent protection, as well as the influence of government policies such as price regulation, state procurement of drugs, and health and safety regulation. Rising research and development (R&D) expenditures in the face of health care cost containment pressures and apparently slowing research productivity give pharma- ceutical companies a powerful incentive to seek out cost savings and new models for innovation, and to the extent that “offshoring” can raise research productivity it will generate substantial benefits to U.S. consumers and taxpayers. However, although we see some evidence of cost-driven geographic redistribution of R&D into new low-cost locations, this process has thus far been limited. 207 

208 INNOVATION IN GLOBAL INDUSTRIES Global Distribution of Activity Many, though not all, new pharmaceutical products are marketed world- wide. “Blockbuster” products—the relatively small fraction of drugs that realize global sales in the range of billions of dollars per year—are normally sold in most middle- and high-income countries. Patients in Organisation for Economic Co-operation and Development (OECD) countries ultimately have access to 80- 90 percent of these products, albeit with longer delays in countries with more stringent price controls or weaker intellectual property (IP) protection. “Minor” products—those products with global sales below $1 billion per year—are typi- cally launched in fewer countries; patients in the average country ultimately gain access to well under half of the total number of drugs introduced worldwide within 10 years of their first sale. New drugs are significantly less likely to be launched in poorer countries and, even if they do ultimately become available, it can take many years. Some countries are noticeably different in these respects; for example, Japan and Italy have much higher frequencies of single-country products. Country-specific regulatory requirements and differences in health care delivery systems may require drug companies to make significant invest- ments in local capacity in regulatory affairs and sales and marketing, even where promotion of pharmaceutical products is highly restricted. Drug manufacture is also a multinational phenomenon, with an active global trade in intermediates (specialty chemicals), active pharmaceutical ingredients, and finished products. Stringent regulatory requirements for manufacturing im- posed by government agencies in major markets such as the United States have extended quality standards worldwide, and several countries have become major loci of manufacturing activity that supplies global markets, notably Ireland and Puerto Rico, as well Israel and India for generic products. R&D, by contrast, is much more geographically concentrated; the bulk of all R&D expenditure occurs in the United States, a handful of European countries, and Japan. Table 1 provides one measure of the global allocation of aggregate industrial pharmaceutical R&D expenditures. Unfortunately, reliable nationally comparable statistics on R&D spending in pharmaceuticals are difficult to ob- tain. National trade associations for the industry often report global rather than national spending by their members, and use varying definitions of R&D. For smaller countries, particularly emerging economies, data are simply unavailable, intermittently available, or of very questionable quality. With these caveats, ag- gregate statistics based on government censuses may nonetheless be informative, and they suggest a relatively stable share of the allocation of total industry R&D expenditure among developed countries, though even these numbers are difficult to compare due to differences in industry definitions, reporting standards, and data collection methods as well as exchange rate issues.   See Lanjouw (2005), Kyle (2006, 2007), and Danzon et al. (2003). 

PHARMACEUTICALS 209 TABLE 1  Business Expenditure on Pharmaceutical R&D by Country 1990 1995 2000 2004 Total BERD at PPP (current million $), of which: 16,853 24,587 33,781 46,216 USA 37.3% 41.5% 38.3% 36.5% EU15 39.8% 36.3% 40.4% 39.0% UK 12.1% 11.8% 13.3% 11.1% France 6.4% 8.5% 7.8% 7.6% Germany 8.1% 5.0% 6.7% 7.5% Italy 5.5% 2.5% 1.9% 1.5% Sweden 2.1% 2.7% 3.7% 3.6% Japan 16.2% 14.9% 14.3% 14.8% Other developed countriesa 6.7% 6.3% 5.8% 8.0% “New Europe”b —. 0.8% 0.9% 1.2% Other emerging economiesc —. 0.1% 0.4% 0.6% NOTES: In 2004, data for Australia, France, Greece, Japan, Mexico, Sweden, and Turkey are inferred from 2003 values and the average annual growth rate (AAGR) of BERD over the past 5 years in that country. In 2003 data for Austria, Denmark, Greece, and Iceland are inferred from adjacent year values and the 5-year AAGR of BERD in that country. The same applies to Austria in 1990 and 1995, and Belgium in 1990. Data for Switzerland may not be consistent over time. aAustralia, Canada, Iceland, Korea, Norway, Singapore, and Switzerland. bCzech Republic, Hungary, Poland, and Slovenia. cTaiwan, Mexico, and Turkey. SOURCES: OECD Main Science and Technology Indicators Vol. 2006 release 02, and UK Pharma- ceutical Industry Competitiveness Task Force: Competitiveness and Performance Indicators 2005. Data published by OECD on business expenditure on R&D in pharmaceuti- cals are one basis for examining the global distribution of research effort in the industry. These data are converted to U.S. dollars using purchasing power parity (PPP) exchange rates and are based on the geographical location of spending rather than the “nationality” of the parent corporation. As shown in Table 1, there was almost a threefold increase in total nominal R&D between 1990 and 2004. However, shares by country or region were relatively constant, with both the United States and the European Union (EU) countries accounting for about 40 percent each of “world” R&D expenditure, Japan about 15 percent, and other developed countries (principally Switzerland) about 7 percent. The emerging economies and former Soviet Bloc countries present in this database have a small but increasing share of this total. It is important to note, however, that there is some R&D activity in countries that are not included here. India and China may account for as much an additional $1.5 billion or more of R&D spending in 2004, and other indicators suggest small, though rapidly growing, levels of   OECD reports over $1 billion in pharmaceutical R&D in China in 2000 but other sources suggest much lower figures. The China Statistics Yearbook on High Technology Industry 2006 published by China National Bureau of Statistics, National Committee of Development and Reform, Ministry 

210 INNOVATION IN GLOBAL INDUSTRIES activity in the Russian Federation and other Eastern European countries such as Romania and the Slovak Republic, in some parts of Latin America, as well as countries in Southeast Asia. Note that Table 1 excludes contributions to the development of new drugs from publicly financed R&D and research supported by private nonprofit orga- nizations. Noncommercial R&D is critically important to the industry and is a major source of “upstream” technology in the form of knowledge externalities from basic research, or spin-off products and entrepreneurial companies. Inter- nationally comparable data on these forms of R&D expenditure are not available for many countries; however, data collected by the OECD suggest that medical sciences account for 20 to 30 percent of academic R&D expenditure in most de- veloped countries. Given that the United States accounts for at least one-third of global public-sector research, it seems clear that the U.S. share of the total global research expenditure in pharmaceuticals from all sources would be significantly higher. Focusing on U.S. pharmaceutical companies, self-reported data compiled by PhRMA, the U.S. trade association, indicates that a significant share of R&D spending by U.S.-based companies was incurred outside the United States. (PhRMA members are largely though not exclusively U.S.-headquartered com- panies.) In 2005, these companies spent just under $9 billion outside the United States (21.5 percent of their total R&D spending), almost all of which was in Western Europe and Scandinavia. Table 2 summarizes these data. Interestingly, the ex-U.S. share of total R&D spending has been quite stable: while total non-U.S. R&D spending reported by these companies increased by 633 percent in real terms between 1980 and 2005, over this period the “abroad” share fluctuated between 17 and 22 percent, with no obvious trend over time. Innovation in the Pharmaceutical Value Chain The pharmaceutical value chain encompasses many activities, ranging from basic scientific research to marketing and distribution. Innovation in the industry is tightly linked to basic biomedical science, and many companies participate actively in basic scientific research that generates new fundamental knowledge, data, and methods. R&D activity is conventionally divided into two phases: of Science and Technology, for example, reports just under 1.4 billion yuan of intramural R&D in pharmaceuticals in 2000—about $700 million at PPP. (However, more than 20 percent of this was for “processing of traditional Chinese herbal medicine.”) The same source gives total pharmaceutical R&D in 2005 at just under 4 billion yuan, a remarkable increase.   2004, for example, U.S. expenditure on academic R&D was more than $42 billion out of a total In for OECD member countries of $125 billion. In addition, the United States has one of the highest shares of medical sciences in total academic R&D. SOURCE: OECD Main Science and Technology Indicators database.   Comparable data are not available for non-U.S.-based companies. 

PHARMACEUTICALS 211 TABLE 2  Ex-U.S. R&D Spending by PhRMA Members, 2005 Area $ Million Share (%) Africa 28.0 0.3 Canada 479.3 5.4 Latin America and Caribbean 174.9 2.0 Asia-Pacific (except Japan) 117.5 1.3 India and Pakistan 10.9 0.1 Japan 1,025.4 11.5 Australia and New Zealand 144.6 1.6 Europe 6,524.7 73.4 Central and Eastern European nations, including Russia 244.6 2.8 Middle East 37.7 0.4 Uncategorized 101.3 1.1 Total 8,888.9 100.0 SOURCE: PhRMA Profile 2007, Table 6. discovery and development. Drug discovery includes basic science and research on disease physiology, identification and validation of “druggable targets” in the body where therapeutic molecules may affect disease processes, identification and optimization of drug candidates, and preclinical testing. The development phase of research focuses on testing in humans, from the first small-scale trials directed at establishing basic physiological data in healthy volunteers through to large-scale trials on patients with the disease, which are designed to provide data on safety and efficacy to support applications for regulatory approval of the drug. Following marketing approval, research often continues to develop improved formulations of the product and to establish safety and efficacy in treatment of additional diseases or patient populations. Reflecting extraordinary advances in biology since the 1970s, the industry has become progressively more science- intensive, relying closely on fundamental advances in physiology, biochemis- try, and molecular biology rather than “brute force” application of large-scale resources. If anything, this process has accelerated over the past decade as the industry has focused on complex and systemic diseases such as cancer, autoim- mune diseases, and psychiatric conditions. Particularly in drug discovery, indus- trial and publicly funded research efforts are deeply intertwined.  Though the bulk of innovative activity is concentrated in drug discovery and development, some R&D is also directed at manufacturing technologies and process improvement. However, stringent regulation of manufacturing processes inhibits experimentation and innovation, and for many drugs manufacturing costs are a small fraction of sales. This limits returns to investing in process innovation   See Gambardella (1995) and Cockburn and Henderson (1998). 

212 INNOVATION IN GLOBAL INDUSTRIES as opposed to investing in developing new products. In some cases advances in manufacturing technology can be important to allow use of advanced delivery systems or new formulations, but for many products innovation in manufacturing and production processes is largely confined to generic producers who compete fundamentally on costs. By contrast, for other drugs, particularly the “large- molecule” therapeutic proteins based on biotechnology, manufacturing costs are substantial and production processes are more tightly linked to research activity (Grabowski et al., Forthcoming). While the industry continues to be dominated by large integrated firms that conduct much of this innovative activity in-house, recent decades have seen significant vertical restructuring of the industry and these firms increasingly rely on externally sourced R&D in both the discovery and the development phases of research. In drug discovery, an active entrepreneurial sector that bridges academic and publicly funded research and industrial science has become a very important supplier of drug candidates and tools for performing R&D. In the development phase, specialist firms (contract research organizations) now play a significant role in conducting clinical trials on behalf of the sponsor of a drug. The causes of this restructuring of R&D activity are complex, ranging from changes in patent law and practice that have extended exclusionary IP rights into “upstream” sci- ence, financial market innovations that have eased access to capital for early-stage companies, and the development of institutions that have encouraged universities and public laboratories to actively promote commercialization.  One consequence of these changes is that pharmaceutical innovation now relies heavily on a complex web of contractual agreements linking a variety of actors at various stages of the drug development process. Danzon et al. (2003) found that over one-third of new drugs approved between 1963 and 1999 origi- nated in alliances between industry participants. Data on strategic technology al- liances show an explosion of collaborative activity in the biomedical sector since the early 1990s. While the total number of new industrial technology alliances captured per year in the CATI-MERIT database grew by 76 percent between 1990 and 2003, the number of new alliances per year in “biotechnology”—which likely captures much of the external sourcing of drug discovery by pharmaceutical com- panies—increased by 818 percent. Furthermore, many of these alliances spanned national boundaries: between 1990 and 2003, 45 percent of the alliances in this database in which U.S.-headquartered companies participated were “U.S.-only”   This is not to say that process innovation cannot generate large aggregate savings for the industry. Macher and Nickerson’s (2006) benchmarking study of pharmaceutical manufacturing suggests that adoption of process-analytical technologies, greater use of IT, and complementary work practices could generate cost savings of up to $50 billion per year.   For a lengthier discussion, see Cockburn (2004).   See National Science Foundation Science Indicators 2006, Table AT04-37. Note that many of these alliances may be relatively short-lived, and the “coverage” of alliance formation in this database is difficult to assess. 

PHARMACEUTICALS 213 (i.e., all participants had U.S. headquarters), 23 percent of the alliances in which European-headquartered companies participated were Europe-only, and only 8 percent of the participated alliances in which Japan headquartered companies were “domestic.” Forces Driving Location of Innovative Activity Pharmaceutical companies have always been able to operate R&D facilities largely independently from other activities: though a typical large pharmaceuti- cal firm operates as an integrated economic entity, it normally conducts R&D in multiple locations around the world. The nature of the product development process, along with historically strong IP rights, and relatively straightforward licensing practices, has allowed pharmaceutical companies to “decouple” manu- facturing and marketing from R&D. This has been the case for many decades, but increased vertical dis-integration in R&D activities since the mid-1980s has further relaxed organizational constraints on the location of research activity, permitting extensive geographic reorganization of R&D across countries and regions as well as vertical reorganization within firms. In the United States, for example, “upstream” firms specializing in new technologies for drug discovery are now often located in different locations (such as Boston and the San Francisco Bay area) than those historically used by the “big pharma” firms concentrated in Philadelphia, New Jersey, Connecticut, and the Midwest.10 Many factors drive these R&D location decisions, and the observed geo- graphical distribution of research reflects complex trade-offs among them. One the one hand, economies of scale and scope in performing R&D, the presence of internal knowledge spillovers, and costs of coordinating activity across dis- persed units suggest that, all else equal, firms should limit geographic dispersion of R&D. Furthermore, some locations may be more intrinsically economically attractive because of lower costs, access to government subsidies, or favorable tax treatment of R&D. Proximity to centers of academic excellence and other forms of noncommercial research also appears to convey benefits such as raised research productivity (see Furman et al., Forthcoming). On the other hand, these economic factors, which tend to concentrate R&D, are offset by political consid- erations. In some countries, pharmaceutical companies face strong political pres- sure to maintain domestic R&D. Some countries, such as the United Kingdom, have explicitly linked the stringency of price regulation to local R&D spending   These marked differences in alliance activity are presumably driven by the relative concentration of potential partners (i.e., biotech companies in the United States). 10  is unclear whether the extent of vertical specialization and accompanying geographic realloca- It tion of effort are more or less pronounced in the United States than elsewhere. The United States has attracted the lion’s share of investment in new enterprises in biotechnology and pharmaceuticals, but there is no evidence that U.S.-headquartered firms rely more or less heavily on external R&D than, say, those headquartered in Europe. 

214 INNOVATION IN GLOBAL INDUSTRIES levels; in other cases, such as in Canada, local R&D spending reflects a political bargain to avoid compulsory licensing. Historically, the United States has been perceived by the industry as a very attractive location for pharmaceutical R&D because of its very limited use of price regulation and government purchasing, and its strong patent rights. 11 In contrast, in the late 1990s, EU governments became very concerned that overly aggressive price controls and hard bargaining by state purchasers were driving away investment in pharmaceutical R&D and adversely affecting the competi- tiveness of EU-based companies, though there is little evidence (see Table 1) of any major shift in R&D spending away from Europe. Episodes such as Canada’s experience with compulsory licensing of pharmaceuticals in the 1970s and 1980s, or more recent examples such as the periodic heated disputes between OECD- based companies and governments of developing countries over pricing of an- tiretroviral drugs, suggest that R&D location decisions can be quite sensitive to government policies directed at lowering the cost of acquiring pharmaceuticals. Notwithstanding its long tradition of excellence in medical and pharmaceutical research, and substantial historical investments by multinational drug companies, Canada experienced a steep decline in domestic R&D activity in pharmaceuti- cals when it introduced its compulsory licensing regime. Only when full patent rights were restored, and a relatively loose drug price regulation scheme was instituted, did commercial R&D spending return to previous levels. Countries such as Australia, which have relatively stringent drug price controls, continue to face major challenges in attracting significant R&D investment by multinational drug companies, in spite of strong academic research capabilities, an attractive business environment, and substantial public support of commercial biomedical research. Beyond these “price” drivers, several other factors have been identified as influencing R&D location decisions. These often work through indirect, or un- priced, effects such as knowledge spillovers that are conveyed by “open” publi- cations, geographic proximity, or communication through informal professional networks rather through economic transactions. For example, drug discovery labs sites tend to specialize in therapeutic areas or scientific disciplines 12 and, since 11 The Hatch-Waxman Act lowers the costs of generic entry and provides incentives to challenge pharmaceutical patents, but it also provides certain protections to patent holders. The United States also has provisions that extend the duration of pharmaceutical patents to offset time lost waiting for regulatory approval. 12 For example, in 2000 Hoffman La Roche operated six major drug discovery facilities: Kamakura, Japan; Penzberg, Germany; Basel, Switzerland; Welwyn Garden City, United Kingdom; Nutley, New Jersey; and Palo Alto, California. A research center in Basel focused on basic research in genom- ics, proteomics, and bioinformatics, while the Kamakura, Penzberg, and Nutley labs specialized in oncology, Basel and Nutley in metabolic disorders and vascular diseases, Basel and Palo Alto in central nervous system disorders, Welwyn Garden City in virology, and Palo Alto in inflammation and genitourinary diseases. 

PHARMACEUTICALS 215 proximity to publicly funded science appears to be an important determinant of research productivity, these often reflect local academic centers of excellence in particular fields. Furman et al. (Forthcoming) show that patenting by pharma- ceutical companies is positively correlated with the volume of academic publica- tions by “local” public-sector scientists.13 The very substantial levels of publicly funded biomedical research in the United States, the United Kingdom, and some other countries has therefore played an important role in sustaining similarly high levels of commercial investment in drug discovery in these countries. More generally, like other knowledge-intensive activities, discovery research appears to display substantial agglomeration externalities. Drug discovery activ- ity tends to “cluster” in a small number of locations around the world: many ma- jor discovery labs are located in the New York/New Jersey/Connecticut standard metropolitan statistical area, Boston, the San Francisco Bay area, the suburbs of Philadelphia, Research Triangle in North Carolina, the Rhine Valley, the suburbs of London, Stockholm, and Tokyo/Kansei. These are conspicuously not low-cost locations, so this clustering suggests substantial offsetting economic benefits derived from being co-located with other firms. Beyond the role of localized knowledge spillovers, benefits from co-location with other pharmaceutical firms include access to skilled labor and “infrastructure” in the form of specialized services and suppliers, and efficient interaction with collaboration partners. 14 A final factor that may affect R&D location decisions is the strength of IP protection. Though there is no obvious connection between the degree of patent protection in the local product market and the productivity of R&D conducted in any given country, the nature of a country’s IP regime appears to affect mul- tinationals’ willingness to conduct R&D activities there. This may be because weak patent protection for products often correlates with weak legal protection of other forms of IP such as trade secrets and associated contractual agreements with employees and suppliers, and limited avenues to enforce these rights. Both patent and nonpatent protection of IP play an important role in maintaining exclusive access to, and control over, proprietary knowledge, and in countries with weak IP companies may have well-founded concerns about “leakage” of valuable information to local competitors. Zhao (Forthcoming) argues that weak 13  To some degree, drug discovery labs operated by different firms within the same region appear to specialize in particular therapeutic classes or scientific disciplines. Cockburn et al. (2002) report, for example, that for their sample of firms commercial R&D in New Jersey was primarily, though not exclusively, focused on cardiovascular therapies, whereas that conducted in the suburbs of London was primarily in antipsychotics. 14  Returning to the example of Hoffman La Roche in 2000, of nine important collaborators identi- fied by the company, Tularik, Affymetrix, Clontech, and Incyte (plus its majority-owned subsidiary Genentech) were geographically proximate to its Palo Alto lab, CuraGen and Progenics were close to its Nutley lab, and Vernalis, Imperial College, and Oxford University were close to Welwyn Garden City. In a study of cross-regional collaborations by a broad set of biomedical technologies, Zhao and Islam (2007) document increasing geographic dispersion of R&D activities of large firms, but also increased internalization of knowledge spillovers within these firms. 

216 INNOVATION IN GLOBAL INDUSTRIES IP regimes need not deter R&D investment by multinationals: absent strong IP rights, companies can nonetheless develop alternative mechanisms for realizing returns on innovation and IP. These mechanisms include rapid “internalization” of knowledge through efficient internal organizational processes and control of complementary assets and may make it possible to profitably exploit low prices of R&D inputs and underutilized domestic innovation capabilities. However, this argument is most appealing for technologies that have a substantial tacit compo- nent, are strongly complementary to other protected assets held by the firm, and have rapid development cycles. This is not the case for pharmaceutical R&D, where results from R&D are often easy to “externalize” and imitate, and product life cycles are measured in decades. Not surprisingly, therefore, R&D activity in pharmaceuticals has histori- cally been concentrated in countries with strong and enforceable IP and has only just begun to grow in countries that have recently adopted OECD-style patent systems under the provisions of the Trade-Related Aspects of Intellec- tual Propoerty Rights (TRIPS) agreement. Compliance with TRIPS requires all World Trade Organization (WTO) members to (ultimately) adopt key features of the patent systems of wealthy industrialized countries, such as a 20-year term, nondiscrimination across fields of technology and nationality of applicants, and effective enforcement procedures. Strong patent protection for pharmaceuticals is controversial in many of these countries (see discussion of India which follows), and the degree to which domestic political pressures will limit the enforceability of patents, or push the limits of the TRIPS agreement by, for example, institut- ing compulsory licensing of drugs, remains to be seen. Patent rights obtained by multinationals in countries such as India give these companies the ability to exclude generics and to set prices above marginal cost. But patents also provide protection for domestic firms conducting R&D, and political choices to weaken or limit patent protection on the products of multinationals may have serious consequences for nascent research sectors in these economies. Impact of Industry Restructuring on Innovation Structural change in the pharmaceutical industry has given pharmaceutical companies more opportunities and much greater flexibility to improve R&D per- formance by reallocating R&D effort between internal and external projects, and across different locations both within countries and around the world. Whether greater globalization of R&D has been caused by this vertical disaggregation of the industry—or vice versa—is an open question. Clearly the two phenomena are closely linked and, beyond the industry-specific factors that have driven vertical disaggregation discussed earlier, more general phenomena affecting many indus- tries (such as improvements in communication technologies, greater international mobility of labor and capital, innovation in capital markets, and international harmonization of IP rules following the TRIPS agreement) have also played a 

PHARMACEUTICALS 217 role in creating new opportunities for collaborative R&D and specialist providers of R&D inputs. There are many reasons to believe that industry restructuring and globaliza- tion may generate substantial gains in R&D performance. R&D outcomes depend critically on resource allocation (which projects to pursue, how much to spend, which to shut down), which at one time was done almost entirely through internal decisions of large vertically integrated firms. In today’s industry, market transac- tions and the price mechanism play a much greater role in resource allocation, with specialization and competition in the supply of research inputs and services. Capital markets play an important role in pricing risk and provide high-powered incentives to entrepreneurial firms, and strong IP rights support a global “market for technology.” These powerful economic forces may well result in significantly faster/better/cheaper drug development. On the other hand, industry restructuring and globalization may be respon- sible for some inefficiencies that limit any gains in R&D performance. There is no guarantee that the market for technology (i.e., licensing and collabora- tion deals) creates reliable price signals, and market-driven resource allocation may therefore generate worse results than those obtained by the internal capital markets of large firms.15 The struggle between entrants and incumbents in the industry may also be wasting significant resources in bargaining costs; payments to intermediaries such as lawyers and bankers; extra organizational overhead dedicated to seeking out, structuring, and operating collaborative ventures; or in defensive investments to improve bargaining and so forth.16 While the ultimate impact of restructuring and globalization on R&D per- formance will surely take decades to become apparent, it is clear that this more open competitive environment presents severe challenges. Heightened competi- tive pressure, greater cost transparency, and global competition have contributed to an extraordinarily high failure rate among would-be entrants to the industry. Of the many thousands of well-financed entrants with strong patent portfolios and exciting science that have attempted to gain a foothold in the industry as a supplier of technology or competitor to the established multinational incumbent firms, only a few hundred have survived. The prospects for new players in the industry based in emerging countries are therefore mixed. Success in the global pharmaceutical industry requires (among other things) substantial and sustained investments in R&D capacity, IP portfolios, and access to leading-edge science. It will likely take many years before new competitors appear on the world stage to present a serious head-to-head challenge to existing OECD-based firms. 15  Markets for “knowledge goods” have long been understood to be subject to numerous forms of market failure such as imperfect and asymmetric information, nonexcludability, limited numbers of buyers and sellers, exernalities, and so forth. 16  See Cockburn (2004, 2007) for further discussion. 

218 INNOVATION IN GLOBAL INDUSTRIES Evidence on Offshoring of Pharmaceutical R&D Rapid growth in technological capabilities in low-cost emerging economies is presenting new opportunities and challenges for pharmaceutical companies. Some geographic redistribution of R&D activity does appear to be taking place. On the one hand, companies located in countries such as India and China are performing more in-house R&D oriented toward developing new drugs, rather than reverse-engineering existing products or improving production efficiency. On the other hand, reflecting the general trend in the industry toward greater specialization and external sourcing of R&D services, OECD-based companies are beginning to look to low-cost countries as suppliers of contract research ser- vices, and growing numbers of clinical trials are being conducted in emerging economies. India and China are the two countries most frequently mentioned in this regard; however, by some indicators significant growth in activity also ap- pears to be taking place in some Eastern European countries, Argentina, Brazil, Taiwan, South Africa, and Israel. Data on this activity are limited. As discussed earlier, internationally and in- tertemporally consistent aggregate statistics on R&D expenditure in this industry are often not available for low-cost locations, and their reliability is difficult to assess. The experience of specific countries may nonetheless be informative, and the case of India is presented in the next section. Case Study: India’s Pharmaceutical Industry Indian pharmaceutical companies have attracted much attention. Chaudhuri (2005) discusses the development of the Indian pharmaceutical industry in detail, charting the rise of companies such as Ranbaxy and Dr. Reddy to their current po- sition as significant global players in generic drug manufacturing. Many of these companies have developed advanced capabilities in low-cost manufacturing, re- flecting local expertise in chemical engineering and a historically process-oriented patent regime. (The success of these companies also reflects their mastery of the regulatory approval process for generic drugs, and associated litigation, in for- eign markets.) Sustained profitability, together with the introduction of domestic product patents for pharmaceuticals in 2005 as part of the TRIPS agreement, 17 has prompted some of these companies to increase R&D spending and take on drug development projects. Chatterjee (2007) analyzes financial statements of Indian pharmaceutical manufacturers and finds very high growth rates of R&D spending for the 40-50 firms that have obtained U.S. patents and/or Food and Drug Administration (FDA) approval of manufacturing processes, and attract the 17  The TRIPS agreement was reached as part of the Uruguay Round of multilateral trade negotia- tions in the mid-1990s. India and other countries with weak or nonexistent IP regimes agreed to put in place OECD-style patent protection by 2005 as a condition of retaining full participation in the WTO. 

PHARMACEUTICALS 219 attention of domestic stock analysts. These firms increased their R&D-to-sales ratio by a factor of 10 to 20 between 1990 and 2000, and doubled it again between 2000 and 2005 to reach respectable average levels of 5-6 percent of sales. Among these, a handful of “elite” firms have raised R&D spending to 10 percent per year or more in 2005, with compound average growth rates of R&D spending over the past decade of 30-50 percent per year.18 Some of these firms have made significant investments in product-oriented research since the mid-1990s, with some proportion of increased R&D expendi- ture directed toward generating novel compounds to be tested as drug candidates (often referred to as New Chemical Entities [NCEs]). As of early 2006, more than 60 molecules developed by Indian companies were reported to be in late stages of preclinical research, and 16 had reached the clinical trials stage, though none have yet completed Phase III or been approved by the FDA.19 Indian firms have also entered into a small number of highly publicized product development partnerships with multinational R&D-oriented firms: examples include Dr. Red- dy’s outlicensing of diabetes candidates to Novo Nordisk and Novartis in 1997, Glenmark’s 2004 outlicensing agreement with Forest Labs to develop a candidate drug for asthma, and Nicholas Piramal’s 2007 agreement with Eli Lilly to conduct clinical trials and regional marketing for various candidates for treatment of meta- bolic diseases. As with such deals in OECD countries, many of these agreements have been terminated or otherwise failed to generate significant licensing revenue for the Indian partner (Frontline, 2002). Note also that all of these NCEs appear to be “small-molecule” chemistry-based drug candidates rather than the “large- molecule” products characteristic of leading-edge biotechnology research. Indian pharmaceutical companies are also reported to be playing a grow- ing role as contract research providers to research-based multinationals. Indian companies have strong capabilities in medicinal and analytical chemistry, process engineering, and organic synthesis developed originally to support generic manu- facturing. Though the nature of the R&D activities covered by these contract research agreements are unclear, expertise in chemistry-driven R&D activities positions these companies well to support the sourcing company’s core product innovations by developing efficient manufacturing processes, “tweaking” candi- date molecules for better bioavailability, or developing formulations optimized for specific marketing purposes. Indian companies are also developing expertise in toxicology and animal studies to support preclinical research, and total expen- diture on clinical research in India has been estimated as $100 million per year for 2005. Small but significant investments are being made by OECD-based compa- 18  Some caution may be warranted in assessing these data. Accounting standards for defining R&D may vary over time, some of the growth in expenditure may reflect cost inflation, and a certain amount may reflect “window dressing” to attract attention from outside investors or potential joint venture partners. 19  “Pharmaceuticals” Report by Ernst & Young for the India Brand Equity Foundation, 2006. 

220 INNOVATION IN GLOBAL INDUSTRIES nies in captive research facilities located in India and in long-term collaborative discovery partnerships with Indian companies. But though these ventures may signal a new phase in the development of India’s pharmaceutical industry, with an increasingly significant role in global innovation, it is important to note that the scale of this activity is currently very small. To put India’s current scale of activity in drug discovery and development (at most 100 early-stage candidates since 200020) in perspective, it is worth noting that worldwide several thousand molecules enter preclinical research and Phase I trials every year. The total expenditure on pharmaceutical R&D in India from all sources is growing rapidly but is unlikely to exceed $500 million per year in the near future, which is less than 1 percent of expenditure in OECD countries. Notwithstanding substantially cheaper inputs to R&D in India (labor costs for skilled scientists are claimed to be as little as one-seventh of U.S. levels), the scale of India’s R&D effort is still very small. Complying with the TRIPS Agreement, India has now implemented an OECD-style patent system. Effects on domestic prices are as yet unclear, but availability of product patents appears to be increasing the number of drugs available to Indian consumers and decreasing the amount of time elapsed be- tween their first worldwide launch and availability in the Indian market. 21 Patent protection may also be playing a role both in stimulating R&D by domestic firms and in supporting multinational companies’ participation in contract research agreements and licensing deals. But drug patents are politically controversial in India, and it remains to be seen whether the operation of the Indian Patent Of- fice and enforcement in domestic courts will provide adequate IP protection for product innovators.22 Other Evidence on the Scope and Volume of Global R&D Though comprehensive and reliable data on global R&D expenditure are not available, the location of discovery and development activity can be tracked using proxy indicators such as patent applications, academic publications, or databases that document clinical trial sites. 20  HSBC Securities, cited in IBEF (2006). 21  See Lanjouw (2007). 22  Novartis is currently fighting a closely watched legal battle to secure Indian patent rights on Gleevec, its breakthrough cancer drug. Novartis’ application for patent rights in India was rejected by the Indian Patent Office in 2006. “The patent office ruling in the Gleevec case has sent two far-reach- ing signals on the new TRIPS-compliant law: First, that India will not grant product patent [sic] for any drug unless it was invented after January 1, 1995. Second, that the standards set by the Indian law for grant of product patent for drugs could be more exacting than even those of advanced countries” (Times of India, February 24, 2006). 

PHARMACEUTICALS 221 Discovery Patent applications provide some information on the location of drug discov- ery activity. Because the United States is largest single pharmaceutical market, U.S. patent protection is likely to be sought for most promising drug candidates. Therefore, both the location of the patent assignee and the address of the inven- tors listed on U.S. patents are useful indicators of the location of drug discovery activity.23 Table 3 shows the geographic distribution over time of granted U.S. applications for pharmaceutical patents assigned to corporations. These data are not comprehensive, since they cover only a single class of patents (IPC A61K), and the sample excludes patents where the country of the assignee cannot be de- termined. Data are tabulated by the date of application, which induces truncation of the sample in later years to due the application/grant lag. Patenting in this sample is dominated by U.S. and European companies. Although the number of patents in this sample that are assigned to corporations based in India and China has grown quite rapidly, as the table shows this growth is from such a small base that the share of companies based in these countries remains very small. Table 4 shows results from a slightly different exercise, breaking down the regional share of drug discovery activity based on the location of the inventors listed on drug patents. Again, although the volume of activity in India and China has grown very rapidly since 1990, it still represents a tiny share of total activity. In 1990, for example, France alone accounted for 661 instances of “inventor- ship,” whereas India had 20 and China had 22. These indicators provide little evidence of substantial global relocation of activity in drug discovery. As more current data become available, we are likely to see substantially increased levels of activity in countries like India and China. But although activity in these countries is growing quite quickly, it will be many years before the share of the locations where drug discovery has traditionally been concentrated is materially affected. Development A somewhat different picture emerges, however, for clinical development. For this aspect of innovative activity in pharmaceuticals, the location of clinical trial sites provides some insight into the global distribution of activity. Berndt, Cockburn, and Thiers (2007) report results from tabulating the location of more than 65,000 trial sites participating in the clinical trials that have been registered on clinicaltrials.gov since 2001. Table 5 summarizes this measure of clinical development activity. Because registration is not compulsory for all trials, the 23  “Home bias” may inflate the U.S. share, and language barriers or lack of experience with the U.S. patent system may result in underrepresentation of some countries. 

222 INNOVATION IN GLOBAL INDUSTRIES TABLE 3  Location of Corporations Obtaining U.S. Pharmaceutical Patents Application Year 1990 1995 2000 2002 Total number of patents in sample 3,414 9,277 7,073 3,257 Regional share of patents based on location of assignee (%) United States 55.1 63.3 58.3 57.2 EU15 24.6 21.7 22.3 22.8 Japan 15.3 7.5 9.0 9.5 Other OECD 2.8 5.0 6.2 6.5 India 0.0 0.1 0.7 1.3 China 0.1 0.0 0.1 0.2 NOTES: Table entries based on the count of U.S. patents in IPC class A61K assigned to corpora- tions whose country can be identified. SOURCE: Author’s calculations based on U.S. Patent and Trademark Office (USPTO) data. TABLE 4  Location of Inventors on U.S. Pharmaceutical Patents Application Year 1990 1995 2000 Total pharmaceutical inventorshipsa 10,582 30,135b 23,923 Regional share (%) United States 42.8 56.7 54.4 EU15 28.6 25.0 24.8 Japan 21.2 10.4 10.7 Other OECD 3.1 5.3 6.0 India 0.2 0.2 1.1 China 0.2 0.1 0.2 Other 4.0 3.0 2.8 aAn inventorship is an instance of an inventor being listed on a patent application; therefore, a single patent with three inventors will generate three observations. b1995 saw a surge of applications at the USPTO in order to secure various procedural advantages before the passage of patent reform legislation. SOURCE: Author’s calculations based on USPTO data. Table entries are based on the number of instances of inventorship for patents falling in IPC class A61K, by country of the inventor and date of application for the patent. TABLE 5  Clinical Trial Sites by Region Year 2000 2001 2002 2003 2004 2005 2006 Total sites 2,385 4,139 6,677 8,034 14,224 23,536 33,045 Share of (%) United States 51.4 48.8 50.9 49.1 54.6 49.5 45.2 EU15 32.2 29.4 28.9 25.7 23.1 26.3 26.8 Japan 0.0 0.0 1.6 2.1 2.0 2.5 2.7 Other OECD 7.5 8.2 10.3 8.0 7.4 8.0 7.2 India 0.0 0.1 0.3 0.8 0.7 0.9 1.0 China 0.0 0.4 0.2 0.5 0.3 0.5 0.9 NOTES: Table entries are based on the number of sites participating in clinical trials registered on clinicaltrials.gov. The average number of sites per trial is 7.6. 

PHARMACEUTICALS 223 extent to which these data are representative of all trials is unclear. (In 1997 the FDA began requiring ex ante registration of trials for life-threatening diseases, and community norms have encouraged participation in trial registries, but it was only when a consortium of editors of the major medical journals [ICMJE] agreed in 2004 to require ex ante registration of trials as a condition of publication that the volume of registrations appears to have begun to approach full coverage.) Growth in the total number of trials, particularly in the early years of this sample, largely reflects increases in the coverage rate rather than increases in the volume of activity. But provided nonregistration does not vary systematically across countries, shares of activity are nonetheless a reasonable metric for the extent of a country or region’s involvement in clinical research. As Table 5 shows, the United States, the EU, and Japan continue to account for the bulk of clinical trial activity, but India and China had a significant and rapidly growing share of activity. Overall, emerging economies and low-cost locations are a relatively small share of global activity, but this is changing rapidly. Table 6 shows the share of global trial sites by geopolitical region and by traditional versus emerging coun- tries. In 2002, more than 90 percent of trial sites were located in “traditional” countries (North America, Western Europe, and Scandinavia), but this proportion has fallen rapidly in recent years, with the share of nontraditional countries in the total number of trial sites rising from 7 to 17 percent. Growth in this form of R&D activity has been particularly strong in Eastern Europe and Asia. Between 2003 and 2006, for example, Malaysia, Philippines, Bulgaria, Chile, Turkey, Argen- tina, the Russian Federation, Thailand, Mexico, and Latvia more than quadrupled their share of global trial sites, and India and China’s shares more than tripled. In a regression analysis of factors driving the global allocation of clinical tri- als, Berndt, Cockburn, and Thiers (2006) find that changes in countries’ share of TABLE 6  Regional Share of Worldwide Clinical Trial Sites (%) Year 2002 2003 2004 2005 2006 North America 58.2 54.1 59.8 54.3 49.5 Western Europe 30.6 26.7 24.1 27.4 27.6 Oceania 3.3 4.0 3.2 4.5 4.6 Latin America 1.7 3.5 3.7 3.3 4.3 Eastern Europe 3.8 7.4 4.9 5.9 8.1 Asia 1.1 3.1 2.5 3.2 4.0 Middle East 0.3 0.3 0.7 0.7 0.6 Africa 1.0 0.8 0.9 1.0 1.0 “Traditional” countries 92.4 85.0 87.4 86.4 82.0 “Emerging” countries 7.1 14.2 11.8 12.6 17.1 Others 0.5 0.9 0.9 1.0 0.9 SOURCE: clinicaltrials.gov. 

224 INNOVATION IN GLOBAL INDUSTRIES trial sites were negatively associated with a measure of cost per patient, but posi- tively associated with changes in the strength of patent protection for biomedical inventions. Two- or fourfold differences in the cost per patient of conducting trials in emerging countries rather than in traditional countries thus appear to be driving a substantial expansion of activity in these locations. Interestingly the share of emerging countries in this activity is highest for large, confirmatory tri- als (Phase III of the drug development process) and lowest for small, early-stage trials that are more closely connected with basic biomedical science, suggesting that some types of clinical research are much less strongly influenced by the clustering and proximity to leading-edge academic research effects that drive the location of drug discovery. Turning to IP issues, the positive association found here between changes in patent protection of biomedical inventions and growth in share of global clinical trials may reflect concerns of both multinational and domestic R&D performers about imitation of their products and their ability to appropriate returns from innovation. Even though sales in these countries are only a very small share of the global pharmaceutical market, and their current profitability is therefore unlikely to be a major driver of R&D decisions, it is clear that they have huge potential for future growth in sales. Changes in patent protection may therefore be important as a guarantee of future profitability in much larger markets, attracting R&D, which will realize returns far in the future. One critical connection between local R&D and local future sales may be the role that late-stage trials can play in building future demand—by familiarizing key opinion leaders in the local medical community with a product, the sponsor of the trial may realize higher volumes or better reimbursement prices once the product is launched. One question of great interest is whether participation in global clinical trials has “spillover” effects in the sense of building local capacity to conduct independent clinical research in support of domestic drug development programs. Clearly, sustained participation in this activity will allow clinical researchers to gain experience, credibility, and skills, and to promote development of support- ing infrastructure and services. But it is likely to take considerable time before emerging countries are able to design and conduct complex trials on a routine ba- sis and develop competitive capabilities in “translational medicine”—the bench- to-bedside combination of clinical investigation with basic research that plays a critical role in drug development. This type of R&D activity relies heavily on participation by skilled physicians, who are a very scarce resource in emerging countries relative to medical needs. Substantial expansion of clinical research will require these countries to make significant investments in medical schools and training programs in order to meet increased demand both for routine care for larger and wealthier populations and for an emerging clinical research sector. It is far from clear how fast or extensive this supply response will be, or what will be the impact of rapid economic growth and consequent expansion of health care on the market for physician services. 

PHARMACEUTICALS 225 Manufacturing and Process Innovation While product innovation remains the major focus of the industry, it is worth noting that process innovation capabilities may play an increasingly important role in the future. Generic products account for a very large proportion of drug consumption, and this will likely grow in the future as patents expire for the cur- rent set of “blockbuster” products. Suppliers of these products compete on costs, and Indian manufacturers in particular appear to have acquired world-leading capabilities in developing low-cost manufacturing processes, which positions them to play a dominant role in supplying low-price products to both developed and developing countries. Looking further into the future, new generations of large-molecule biotech drugs will likely displace the current set of chemistry- based products, particularly for diseases such as cancer. But these drugs are no- toriously expensive and difficult to produce, suggesting an important future role for manufacturing innovation that should bring costs down to the point at which these products can be profitably supplied to large lower-income markets. Clusters of activity and development of these manufacturing innovation skills are already occurring in “new” locations such as Singapore, which may become important locations for this type of manufacturing in the same way that they have for other technologies such as semiconductors. Discussion and Conclusions Globalization is not a new phenomenon in pharmaceuticals. At least since the 1950s, innovation in this industry has been geographically dispersed, with phar- maceutical companies conducting R&D in multiple locations around the world. Over the past 25 years, however, gains from specialization in different aspects of the drug discovery and development process, changes in IP rules, easy access to venture capital, and the refinement of collaborative business models have driven greater vertical disaggregation of the industry, along with greater geographic dispersion of R&D. Pharmaceutical companies now have more opportunities and much greater flexibility to reallocate R&D effort between internal and ex- ternal projects and among different locations both within countries and around the world. While the industry continues to be dominated by vertically integrated multinational firms whose activities span the entire chain of research, from basic science to postapproval epidemiology, these firms compete, collaborate, and in- teract with a wide variety of new actors. This interaction is not confined by national boundaries, and the development of high-quality, low-cost research capabilities outside the United States and Eu- rope presents significant opportunities for multinational R&D-based pharmaceuti- cal companies to outsource some aspects of the innovation process. The available evidence suggests that emerging countries are playing a rapidly growing role in the global research effort, albeit currently very small compared to the scale of 

226 INNOVATION IN GLOBAL INDUSTRIES activity in the United States and Europe. In large part, this role has been enabled by structural change in the industry creating opportunities for collaboration and markets for specialized skills and services. But it has also been facilitated by the general trends in the world economy that are driving globalization of many in- dustries, including factors such as improvements in communication technologies, greater international mobility of labor and capital, accumulation of human capital and business infrastructure in low-cost emerging economies, and harmonization of IP rules following the TRIPS agreement. From a U.S. perspective, this “offshoring” of pharmaceutical R&D has potential benefits to be considered as well as the costs from any loss of the stable, high-wage employment characteristic of this industry. The pharmaceutical industry currently faces what some commentators have termed a “productivity crisis.” The number of new drugs approved each year appears to be stagnating, with limited progress made in recent decades in treating some major diseases, yet R&D expenditure has been growing very rapidly. One widely used indicator of research productivity is the cost per new drug approved, accounting for failed projects and the time value of money, which has been rising at an alarming rate. The most recent in a series of studies from the Tufts Center for the Study of Drug Development (Dimasi et al., 2003) estimates the present value of R&D expendi- tures to bring a new drug to market to be $802 million per FDA-approved new drug. In constant year 2000 dollars, this $802 million is more than double the $318 million estimated in an earlier 1991 study, and almost six times larger than the $138 million figure obtained in a 1979 analysis. Recent industry estimates are now well in excess of $1 billion per successful new drug. There are good reasons to believe that this “crisis” may not be as severe as some media accounts suggest. For example, simply counting the number of new drugs approved is not very helpful if the “quality” of each new drug measured in terms of health impact is changing over time.24 Nonetheless, there are serious grounds for concern. To support these continued investments in R&D in the face of these rising costs, pharmaceutical companies need to realize very substantial sales revenues, which may be difficult to sustain in the face of political pressure around the world to restrain increases in health care expenditure. To the extent that the drug development process can be made more efficient through greater flexibility in resource allocation, more experimentation in business models, and greater use of low-cost inputs, offshoring of some aspects of R&D has the wel- come potential to reduce the cost of developing new drugs. Furthermore, at least in the short term, there are limits to the types of activity that are likely to be relocated outside the industry’s traditional locations. Sub- stantial offshoring of R&D activities is most likely to occur where the research activity is relatively routinized, uses large amounts of relatively low-skilled labor, and does not need to be tightly integrated or co-located with other R&D 24  See Cockburn (2007) for a fuller discussion. 

PHARMACEUTICALS 227 activities. Large-scale, late-stage clinical trials with “low-tech” endpoints (such as measuring blood pressure) are examples of this kind of activity, and indeed the global allocation of research effort in these areas shows signs of a significant response to cost differences across countries. As suggested earlier, greater use of, for example, India’s chemistry-oriented R&D capacity presents a “win-win” opportunity for U.S.-based companies to improve R&D efficiency by contract- ing out relatively routine work on process engineering, compound synthesis, and medicinal chemistry while focusing on other aspects of R&D. These other aspects of the innovation process—less routinized, more science- intensive—are much less likely to relocate to low-cost locations. Decisions about where to locate science-intensive drug discovery appear to be much less sensi- tive to labor costs and may be driven primarily by factors such as proximity to leading-edge academic research and “cluster” externalities. The benefits of co- locating R&D labs along with competitors in locations such as Baltimore/Wash- ington, Boston, or the San Francisco Bay area, which feature agglomerations of commercial research, university science, and academic medical centers, coupled with “thick” local markets for specialized inputs and human capital, are very substantial. Any labor cost savings from relocating R&D labs to countries such as India and China are unlikely to compensate for the negative effect of losing access to these benefits. Substantial geographic redistribution of core R&D effort in this industry therefore seems likely to occur only if and when these offshore locations de- velop their own critical mass of academic biomedical science and supporting complementary infrastructure. Some countries, such as Singapore and Taiwan, have put in place major programs to create research infrastructure and attract leading academic researchers, but this will take significant amounts of time and money. Emerging countries such as India and China, which have pockets of aca- demic excellence in biomedicine but have had historically relatively low levels of public support for biomedical research, face an uphill struggle to develop this national capacity. The United States enjoys a dominant position in the world in publicly funded biomedical research. Provided U.S. taxpayers continue to fund this level of support, and public policy sustains the institutions of Open Science that support the productivity and vitality of academic science, the United States seems likely to remain the location of choice for science-intensive pharmaceuti- cal R&D. Open Science, with its curiosity-driven, investigator-initiated agenda and priority- and publication-based incentives, is a distinctive and vital component of the biomedical innovation system. But, particularly in the United States, science has become increasingly “propertized” by the extension of the patent system into basic research and the enthusiastic participation of universities and individual academics in patenting and entrepreneurial activity in life sciences. While this activity has clear benefits in terms of facilitating technology transfer and attract- ing venture-backed investment, it also carries with it less obvious costs in terms 

228 INNOVATION IN GLOBAL INDUSTRIES of weakening the institutions of Open Science, limiting access to research tools and data, and forcing congruence between the agenda of academic research and commercially attractive areas of inquiry. The long-run competitiveness of the U.S. pharmaceutical industry may therefore require careful management by policy makers of conflicts between exclusion-oriented IP rights and traditional academic norms. Patent policy may play a somewhat different role in influencing R&D lo- cation decisions through its effect, in conjunction with price regulation and government purchasing, on domestic returns to pharmaceutical R&D. As dis- cussed earlier, policy efforts to restrain pharmaceutical spending in the United States—for example, through changes to the Medicare drug benefit or reform of the Hatch-Waxman Act—should therefore carefully consider the impact of such measures on R&D location decisions. Of course, government actions elsewhere in the world directed at lowering drug prices may work in the opposite direction. “Unenthusiastic” implementation of the TRIPS patent provisions in emerging countries and unpredictable operation of these new patent systems in practice may cause multinationals to rethink decisions to expand R&D activity in these countries. The closely watched Gleevec case in India has major implications for the future of pharmaceutical R&D in that country, whether performed by domestic competitors on their own account or in partnership with multinationals. Similarly, current efforts by the government of Thailand to force one U.S.-based company (Abbott) to lower the price of some its drugs to Thai consumers are unlikely to encourage multinational companies to engage in R&D activities in Thailand in the future. Widespread actions of this kind may have a significant negative impact on offshoring trends. Major changes in the existing international allocation of innovative effort in the pharmaceutical industry are unlikely, particularly in the short run. Compared to other technologies, this industry moves relatively slowly: very long product development cycles, and necessarily conservative organizational structures and processes imposed by health and safety regulation, make it difficult for pharma- ceutical companies to make large, transformative changes to their business as fast as some firms can in other industries. Recent rapid expansion of research capacity in low-cost emerging countries will benefit U.S.-based multinationals (and U.S. consumers of pharmaceuticals) by lowering R&D costs in some activities, but this rapid expansion is highly unlikely to cause a “tsunami” of professional job losses in pharmaceutical research. In the long run, emerging countries may suc- ceed in developing a large enough base of local academic and publicly funded biomedical science to threaten the substantial competitive advantage that the United States currently enjoys in this area. Commercial R&D location decisions are tightly linked to publicly funded science; therefore, it will be necessary for public policy in the United States to play close attention to vitality and viability of academic- and government-supported biomedical science if the United States is to retain global leadership in the pharmaceutical industry. 

PHARMACEUTICALS 229 ACKNOWLEDGMENTS I thank Ernst Berndt, Jeff Furman, Jeffrey Macher, and David Mowery for helpful discussions and suggestions. REFERENCES Berndt, E., I. Cockburn, and F. Thiers. (2006). Intellectual Property Rights and the Globalization of Clinical Trials for New Medicines. Seminar Presentation, RAND Institution, Santa Monica, November 2006. Berndt, E., I. Cockburn, and F. Thiers. (2007). The Globalization of Clinical Trials for New Medicines into Emerging Economies: Where Are They Going and Why? Conference Paper, UNU-MERIT Conference on Micro Evidence on Innovation in Developing Countries, Maas- tricht, May 31, 2007. Chatterjee, C. (2007). Fundamental Patent Reform and the Private Returns to R&D—The Case of Indian Pharmaceuticals. Manuscript, Heinz School, Carnegie Mellon University. Chaudhuri, S. (2005). The WTO and India’s Pharmaceuticals Industry: Patent Protection TRIPS and Developing Countries. New Delhi: Oxford University Press. Cockburn, I. M. (2004). The changing structure of the pharmaceutical industry. Health Affairs 23(1):10-22. Cockburn, I. M. (2006). Blurred boundaries: Tensions between open scientific resources and com- mercial exploitation of knowledge in biomedical research. Chapter in Advancing Knowledge and the Knowledge Economy, B. Kahin and D. Foray, eds. Cambridge: MIT Press. Cockburn, I. M. (2007). Is the pharmaceutical industry in a productivity crisis? Chapter in Innovation Policy and the Economy, Vol. 7, A. Jaffe, J. Lerner, and S. Stern, eds. Cambridge: MIT Press for the National Bureau of Economic Research. Cockburn, I. M., and R. Henderson. (1998). Absorptive capacity, coauthoring behavior, and the orga- nization of research in drug discovery. Journal of Industrial Economics 46(2):157-182. Cockburn, I. M., J. Furman, R. Henderson, and M. Kyle. (2002). Geographic Location and the Pro- ductivity of Pharmaceutical Research. Conference paper, NBER Summer Institute, July 2002. Danzon, P., Y. R. Wang, and L. L. Wang. (2003). The Impact of Price Regulation on the Launch Delay of New Drugs—Evidence from Twenty-Five Major Markets in the 1990s. NBER Work- ing Paper 9874. Dimasi, J.R. Hansen, and H. Grabowski, 2003. The price of innovation: New estimates of drug de- velopment cost. Journal of Health Economics 22:151-185. Drug trials and questions. (2002). Frontline 19(19). September 14. Furman, J., I. Cockburn, R. Henderson, and M. Kyle. (Forthcoming). Public & private spillovers, location, and the productivity of pharmaceutical research. Annales d’Economie et Statistique. Gambardella, A. (1995). Science and Innovation. Cambridge University Press. Grabowski, H., D. Ridley, and K. Schulman. (Forthcoming). Entry and competition in generic bio- logicals. Working Paper, Fuqua School of Business. Managerial and Decision Economics. IBEF (India Brand Equity Foundation). (2006). Pharmaceuticals. Report by Ernst and Young for the India Brand Equity Foundation/Confederation of Indian Industry/Ministry of Commerce and Industry. Kyle, M. (2006). The role of firm characteristics in pharmaceutical product launches. RAND Journal of Economics 37(3):602-618. Kyle, M. (2007). Pharmaceutical price controls and entry strategies. Review of Economics and Sta- tistics 89(1):88-99. Lanjouw, J. (2005). Patents, Price Controls, and Access to New Drugs: How Policy Affects Global Market Entry. NBER Working Paper 11321. 

230 INNOVATION IN GLOBAL INDUSTRIES Lanjouw, J. (2007). Patents, Price Controls and the Arrival of New Drugs: How Policy Affects In- ternational Launch Patterns. Paper presented at UC Berkeley Lanjouw Memorial Conference, March 2007. Macher, J., and J. Nickerson. (2006). Pharmaceutical Manufacturing Research Project. Available at http://faculty.msb.edu/jtm4/PMRP%20results/. Accessed December 12, 2006. Zhao, M. (Forthcoming). Conducting R&D in countries with weak intellectual property rights protec- tion. Manuscript, University of Minnesota. Management Science. Zhao, M., and M. Islam. (2007). Cross-Regional Ties within Firms: Promoting Knowledge Flow or Discouraging Spillover? Working Paper, Stephen M. Ross School of Business.