the pharmaceutical industry - ScienceDirect.com

Chapter25

THE PHARMACEUTICAL INDUSTRY* F.M. SCHERER HarvardUniversity

Contents Abstract 1. Introduction 2. Distinguishing characteristics 3. Pharmaceutical industry structure 4. Research, development, and the discovery of new drugs 5. Government regulation of new drug introductions 6. Patents and pharmaceutical innovation 6.1. The unusual importance of drug patents 6.2. Consequences of the Uruguay Round agreement

7. Pricing 7.1. Pricing branded drugs 7.2. Tort liability risks and prices 7.3. 7.4. 7.5. 7.6.

Generic drug competition Branded vs. generic drug price competition Stimulating generic substitution New institutions; New power relationships

8. Profits and price controls 9. Conclusion References

*The author thanks Ernst Berndt, Randall Ellis, and Joseph Newhouse for helpful comments. Handbook of Health Economics, Volume 1, Edited by A.J. Culyer and J.P Newhouse © 2000 Elsevier Science B. V All rights reserved

1298 1299 1300 1303 1305 1308 1316 1317 1318 1319 1320 1320 1321 1322 1324 1325 1328 1331 1332

1298

EM. Scherer

Abstract This Handbook chapter surveys the extensive body of research on the economics of the pharmaceutical industry (with peripheral attention paid also to regulated medical devices). Pharmaceuticals is one of the world's most research-intensive industries, generating a continuing steam of new products that save lives and raise the quality of life. The discovery of new drugs has evolved over time from a decidedly empirical process to one based to a considerable degree upon fundamental scientific knowledge. Rich linkages have emerged between profit-seeking manufacturers and basic research performers such as universities and national laboratories. The safety and efficacy of new pharmaceutical products are stringently regulated in most industrialized nations, adding to clinical testing costs. Because of high expenditures on research, development, and clinical testing and because new products, once proven, might be imitated easily, patent protection is unusually important. The extension of patent protection to third-world nations under Uruguay Round Treaty mandates has precipitated vigorous policy debates. Patents, first-mover advantages, and the lack of good substitutes for significant new drugs often give rise to substantial monopoly power, against which many national governments have counterpoised a diverse panoply of price control mechanisms. When patents expire, however, generic substitutes often introduce vigorous price competition. The extent to which generics capture market share from the branded original drugs depends upon government regulatory policies, the reimbursement strategies of health care insurers, and the organization of health care provider institutions. JEL classification:Ill, L5, L65, 031, 034

Ch. 25:

The PharmaceuticalIndustry

1299

1. Introduction This chapter analyzes the supply of important inputs into the provision of health care notably, medicines, vaccines, and other drugs, which will be called "pharmaceuticals" here. Because their supply is affected in analogous ways by regulatory institutions and changes in technology, passing attention will also be devoted to medical devices such as pacemakers, arterial grafts, prosthetics, and the like. Pharmaceutical products account for sizeable but widely varying proportions of total expenditures on health care in the world's industrialized nations. In 1993 or 1994, for instance, their share of total health care spending was 8.3 percent in the United States, 15.4 percent in France, 18.5 percent in Germany, and approximately 29 percent in Japan [PhRMA (1997, p. 48) and Neary (1995, p. 22)]. Some reasons for the observed differences will emerge as our analysis proceeds. Variations in income levels and national health care institutions lead to even larger differences in average annual expenditures per capita on pharmaceutical products among the world's principal regions, as the following United Nations staff estimates for 1990 (calculated in 1980 U.S. dollars at prevailing exchange rates) indicate1: North America European Community Other Western Europe Japan South and East Asia China Latin America Sub-Saharan Africa

$123.90 102.90 85.70 276.60 5.00 4.80 20.30 3.30

The boundaries delineating the pharmaceuticals industry, as it will be analyzed here, are imprecise. An indication of the principal components is provided by U.S. Bureau of the Census data (1997) on the sales (in millions of dollars) of domestic manufacturers in diverse categories for 1995: Finished pharmaceutical preparations for human use Vaccines and other biological products for human use Blood and blood derivatives for human use Diagnostic products (used in vivo and in vitro) Pharmaceutical and biological preparations for veterinary and industrial use Bulk medicinal and botanical products (many used as inputs to other sectors)

48,066 4,986 1,624 8,271 2,601 7,111

The source is Ballance et al. (1992, pp. 30-31). Herbal and other naturally occurring traditional medicines are excluded from the estimates. For 1995 estimates based upon purchasing power parity rather than prevailing exchange rates, and showing France in the lead followed by the United States and Japan, see PhRMA (1998, p. 75).

FM. Scherer

1300

The first category will be emphasized here, although peripheral attention will be devoted to analogous products. Manufactured products other than pharmaceuticals and biologicals used primarily in health care had the following U.S. sales at wholesale (in millions of dollars) during 1995: Surgical, orthopedic, prosthetic, and therapeutic appliances and supplies Other surgical appliances and supplies Electronic hearing aids Surgical and medical instruments and apparatus Dental equipment and supplies Electromedical diagnostic and therapeutic equipment Ophthalmic goods (including eyeglasses)

10,566 964 476 14,285 2,117 6,852 2,715

2. Distinguishing characteristics The pharmaceuticals industry has a number of characteristics differentiating it from most industries serving demands outside the health sector. For one, pharmaceutical products can often be obtained legally only when the consumer presents a prescription from a physician. Consumers in effect cede an important decision-making role to physicians, who include in their decision calculus professional responsibilities and malpractice liability risks and who, usually incurring no obligation to pay for the consumer's purchase, may serve as imperfect agents for the consumer. The prescription system is much more prevalent in industrialized nations than in developing nations [see Peltzman (1987)]. Its mandatory extension in the United States to a wide range of new pharmaceutical products occurred through regulatory agency action in 1938 without a clearly focused Congressional debate [Temin (1979a)]. Under U.S. regulations, paralleled by those of many other nations, a distinction is made between prescription drugs, which require a physician's intervention and which can be purchased for out-patient use only from licensed pharmacists, and "over the counter" or "proprietary" drugs, which can be obtained without prescription from a broader array of retail outlets. In the United States, prescriptions are normally required for new drugs, but chemical entities shown by substantial experience to have minimal toxicity risks and whose use (often at less intense dosage levels) does not require the advice of a physician can be accorded over-the-counter status. See Schachtel (1994) and Weintraub (1994). Since 1972, many drugs such as acetaminophen, ibuprofen, numerous antihistamines, the four leading anti-ulcer agents (Tagamet, Zantac, Pepcid, and Axid) and diverse topical antibiotics and steroids have been shifted to OTC status in the United States. The share of OTC drugs in sales of all U.S. pharmaceutical entities for 1988 was estimated to be 23 percent [Ballance et al. (1992, p. 35)]. Comparable estimates for twelve other nations ranged from six percent (for Italy) to 42 percent (for Saudi Arabia), with an unweighted mean of 18 percent. Second, consumers' purchases of prescription drugs, like their purchases of health care services, are often reimbursed in whole or in part by insurance. As other contri-

Ch. 25:


1301

butions in this volume analyze more fully, with insurance a wedge intrudes between the demand curve derived from consumers' income and the full price paid and the (higher) demand curve reflecting quantities consumed at prices net of insurance payments. This divergence of demand functions leads to reduced demand elasticity and increased purchase of the insured item. See, e.g., Leibowitz et al. (1985) and Newhouse (1993, pp. 165-171). Western European nations have tended to provide more generous insurance coverage of pharmaceutical product purchases than the United States. However, convergence has been occurring. In the U.S., the share of outpatient drug costs paid for out-of-pocket by consumers fell from 82.4 percent in 1970 to 33.9 percent in 1995, largely as a result of expanded private health insurance coverage. See PhRMA (1998, p. 51) and U.S. Office of Technology Assessment (1993, pp. 238-263). Meanwhile consumers' payment shares have been rising in Europe as government health authorities struggled to combat ever-rising costs by requiring substantially increased co-payments. Gone, it is fair to say, are the days the author experienced in Germany during the early 1970s when most consumers obtained prescriptions even for common aspirin so that their health insurance would bear the cost. Even if their outlays are not reimbursed by insurance, consumers, and especially affluent consumers in industrialized nations, are willing to pay a considerable price to combat a painful and debilitating infection, reduce the risk of a heart attack, alleviate the pain of arthritis and muscle tears, or soothe the tensions of a neurotic world. What this means, and especially when decision-making by physicians (which constricts substitution opportunities) and reimbursement by insurers are factored in, is that the demand for many drug products is fairly inelastic up to rather high price levels before income effects begin imparting appreciable elasticity. For instance, Berndt et al. (1995) estimated the demand elasticity for H2 -antagonist anti-ulcer drugs as a group at -0.69 and for individual molecular variants in the range of -0.74 to -1.03. See also (on cephalosporins) Ellison et al. (1997). These demand-side characteristics interact with the presence of monopoly power on the supply side to support prices that commonly exceed drug production costs by a substantial margin. The newest ethical drugs are covered in many nations by patents, which may limit the supply alternatives for a given chemical entity to a single firm. Even when multiple alternatives exist, consumers' limited knowledge of efficacious substitutes and insulation of decision-making physicians from the need to pay for the drugs they prescribe can lead to the persistence of high prices for well-established pharmaceutical products. Where MC is marginal cost, P the product price, and ep is the absolute value of the price elasticity of demand, a monopolist will maximize its profits by setting a price such that the price-cost margin (P - MC)/P = 1/ep. Thus, prices will be raised until demand becomes price-elastic. This condition is inconsistent with evidence of price elasticities below unity, but can be readily reconciled with low observed elasticities if the producers of substitute drugs engage in price-eroding oligopolistic rivalry, or if they exercise restraint to deter rival entry or avoid possibly adverse public relations repercussions from high prices. See Scherer and Ross (1990, pp. 227-229 and Chapter 10).

1302

EM. Scherer

Using Census data, one can approximate the price-cost margins industries realize by computing sales - materials purchase costs - in-plant payroll costs sales See Scherer and Ross (1990, pp. 418-419). Such estimates are probably downwardbiased, since some payroll costs may be fixed and hence not part of true marginal cost. Among 459 four-digit manufacturing industries covered by the U.S. Census in 1987, pharmaceuticals (SIC 2834) had the sixth-highest price/cost margin at 61.4 percent; the average for all manufacturing industries was 30.5 percent. 2 Public perception that drug prices are set monopolistically has triggered numerous governmental interventions into price-setting and other aspects of pharmaceutical marketing. See Comanor (1986) and U.S. Office of Technology Assessment (1993). This is a subject to which we return later. Patents are significant in part because the pharmaceutical industry is extraordinarily dynamic technologically. Among the roughly 230 narrowly-defined U.S. industries for which comparable data have been collected, pharmaceuticals has for at least two decades devoted the largest fraction of its sales receipts to research and development [U.S. Federal Trade Commission (1985, p. 31)]. The intensity with which it pursues research has been rising over time. In 1975, companies with membership in the Pharmaceutical Research and Manufacturers of America association had a weighted average R&D/sales ratio of 11.6 percent in their ethical drug divisions; by 1995, the comparable average was 19.4 percent [PhRlMIA (1997, p. 12)].3 From the industry's R&D efforts has emerged a stream of new pharmaceutical products, some of which have made significant contributions to the quality of health care. Many afflictions that once were untreatable or even fatal can now be combatted with routine success through drug therapy. From an analysis of matched U.S. prescription and hospitalization records for the years 1980 and 1992, Frank Lichtenberg (1996) found that a $1 increase on the purchase of pharmaceuticals was associated on average with a $3.65 reduction in hospitalization expenditures. Innovative drugs had a particularly marked impact in reducing hospitalization costs. A subsequent statistical investigation by Lichtenberg (1998) revealed significant life-extending effects from the introduction of important new drugs. Product differentiation, which affects the drug makers' discretion in setting prices, has been enhanced by vigorous marketing efforts. Under the so-called Dorfman-Steiner 2 Industries with higher PCMs than pharmaceuticals were cigarettes (67.6 percent), breakfast cereals, soft drinks, cosmetics and toilet preparations, and chewing and smoking tobacco. 3 Company divisions selling over-the-counter drugs are much less R&D-intensive. In 1977, the latest year for which comprehensive disaggregated data are available, among companies providing breakdowns for the Federal Trade Commission's Line of Business reporting program (1985, p. 31), the average R&D/sales ratio for prescription drug operations was 10.2 percent, but only 2.9 percent for over-the-counter drug operations. In the surgical and medical instruments category, 1977 R&D outlays were 3.8 percent of sales. The median ratio for 230 reporting lines of business spanning virtually all of the U.S. manufacturing sector was 1.0 percent.

Ch. 25:


1303

theorem [Dorfman and Steiner (1954)], a profit-maximizing firm will seek a ratio of advertising (or other promotional activity) to sales that satisfies the relationship (A/S) = ea[(P - MC)/P], where A is the amount of advertising, S is sales volume, (P - MC)/P is the price-cost margin explained earlier, and ea is the advertising elasticity of demand. 4 Because price-cost margins are relatively high in pharmaceuticals and because physicians and consumers are responsive to promotional activity, 5 appreciable resources are devoted to it. Nevertheless, one finds here striking differences between prescription and over-the-counter drug sales and also among nations. Traditionally, prescription drugs were sold in the United States primarily through "detailing" - that is, through one-on-one encounters between drug makers' field sales representatives and individual physicians. Advertising was confined largely to medical profession journals. However, during the 1990s, and especially after regulatory changes in 1997, companies attempted to cultivate end-consumer demand for their prescription products through advertising in general readership publications. Such direct-to-consumer advertising, which in the United States reached $1 billion during 1998, is prohibited by law in most European nations.6 For over-the-counter drugs, the emphasis both in Europe and the United States is strongly on advertising, and especially (in recent decades) on television advertising. In 1977, OTC sellers in the United States devoted 20.2 percent of their sales receipts to media advertising - the highest such ratio recorded in 225 reporting industries [U.S. Federal Trade Commission (1985, p. 31)]. The corresponding media advertising figure for prescription drug vendors was 4.0 percent (28th in rank among 225 industries). Combining outlays on media advertising and other forms of sales promotion (including detailing, distribution of free samples, and in-store displays), total selling costs in 1977 were 35.6 percent of sales for over-the-counter drug providers and 19.4 percent of sales in prescription drug operations. Total prescription drug advertising and promotion outlays in the U.S. market during 1997 were estimated to be $12 billion, or 18 percent of ethical drug sales [Zuger (1997)].

3. Pharmaceutical industry structure The term "pharmaceutical industry" as used here will refer to the collection of companies that manufacture ethical and over-the-counter drugs. Still one must recognize that other actors are involved in significant ways - notably, the retail pharmacies from which consumers purchase drugs taken outside health care institutions, the health care

4 That is, the percentage by which the quantity demanded shifts for a given percentage change in advertising (or other promotional activity). 5 For estimates of elasticities for "detailing" and two kinds of advertising, see Berndt et al. (1995, p. 103). Price elasticities may in turn be reduced by advertising. See Rizzo (1997). 6 See "Pill pushers", The Economist, August 9, 1997, pp. 58-59, and "Go on, it's good for you", The Economist, August 8, 1998, pp. 51-52.

1304

EM. Scherer

institutions administering in-patient medications, and the wholesalers linking manufacturers with "downstream" outlets. In 1995, members of the Pharmaceutical Research and Manufacturers of America made 78.9 percent of their ethical drug sales in the first instance to wholesalers, 12.1 percent directly to retailers, 4.8 percent to hospitals (who dispensed 14.3 percent of total prescription drugs by dollar value, securing the balance of their supplies from wholesalers), and 4.2 percent to other institutions [PhRMA (1997, pp. 30-31)]. Traditionally, retailers in the United States realized a 40 percent gross margin on the prescription drugs they sold to consumers, which meant that of the average dollar spent by a consumer in 1966, 40 cents went to the retail pharmacist to cover operating costs and profit, nine cents to wholesalers, and 51 cents to manufacturers. Task Force on Prescription Drugs (1968, pp. 54-55). To defend their margins, retail pharmacists were among the strongest supporters of resale price maintenance - i.e., the fixing by manufacturers of minimum prices below which their retailers could not sell. 7 See Scherer (1997). Partly because the support in U.S. law for resale price maintenance was eliminated in 1975, but even more because of events analyzed in a later section, these relationships changed dramatically during the 1980s and 1990s, squeezing the relative gross margins of retailers and wholesalers. Over the same period, chain outlets displaced approximately 15,000 independent retail pharmacies. The manufacturers of drugs, both ethical and OTC, are a diverse lot. The U.S. Census Bureau (1992) reported 640 companies to be operating in the "pharmaceutical preparations" industry (SIC 2834) during 1987. The leading eight companies accounted for 36 percent of total industry sales; the leading 20 companies for 65 percent. Since the early 1980s the U.S. industry has been augmented by hundreds of companies formed to pursue opportunities opened up by the advance of biological science. By the mid-1990s, only a dozen or so of these drug-oriented biotechnology companies had progressed to the point of marketing profitable products. Most of the U.S. pharmaceutical manufacturing leaders are multinational enterprises (MNEs), operating not only in the United States but in many other nations. Incentives to take advantage of unique and exclusive product franchises by making sales throughout the world are so strong that pharmaceuticals is one of the most multinational of industries. Markets outside a pharmaceutical maker's home territory are penetrated by exporting, through the establishment of production facilities overseas, or through licensing of products to companies with already-existing sales channels and regulatory relationships. According to United Nations estimates, pharmaceutical imports averaged 8.2 percent of domestic consumption during 1989 in developed nations and 19.8 percent in less-developed nations [Ballance et al. (1992, p. 52)]. In 1980, approximately 27 percent of the world's pharmaceutical demand was satisfied through local production by foreign-owned companies [Ballance et al. (1992, pp. 68-78)]. The extent of MNE

7 In some continental European nations, retail pharmacies are effectively cartelized, among other things enjoying governmental protection against the formation of retail drug chains or the entry of nearby competing pharmacies unless specified population density thresholds are exceeded.

Ch. 25:

1305


operation is believed to have increased since then. In 1995, members of the Pharmaceutical Research and Manufacturers of America recorded ethical drug sales of $65 billion within the United States and $37 billion outside the United States [PhRMA (1998, p. 100)]. The multinationality of leading pharmaceutical manufacturers increased during the 1980s and 1990s in part through mergers. Among the mergers that joined sizeable companies with disparate national home bases were the following (with the acquiring firm listed first): Beecham (U.K.) - SmithKline Beckman (U.S.) Rhone-Poulenc (France) - Rorer (U.S.) Hoffmann LaRoche (Switzerland) - tGenentech (U.S.) Hoechst (Germany) - Copley (U.S.) Hoffmann LaRoche (Switzerland) - Syntex (U.S.) Bayer (Germany) - Sterling Drug (part) (U.S.) Hoechst-Roussel (Germany) - Marion Merrell Dow (U.S.) Pharmacia (Sweden) - Upjohn (U.S.) Hoffmann LaRoche (Switzerland) - Boehringer Mannheim (Germany)

1989 1990 1990 1993 1994 1994 1995 1995 1997

Other mergers linked companies of considerable scale with home bases in the same nation. These included the fusion of two leading Swiss firms, Ciba-Geigy and Sandoz, to form Novartis in 1996; the merger of two leading British companies, Glaxo and Burroughs Wellcome, in 1996; and, in the United States, the 1989 merger of Bristol-Myers with Squibb, and the 1994 acquisition of American Cyanamid by American Home Products. Altogether, pharmaceutical company acquisitions valued at well over $200 billion occurred between 1989 and 1998. In wholesaling too, a merger wave shrank the number of companies until the leading four firms accounted for 80 percent of U.S. wholesaling activity. Attempts by the four leading wholesalers to merge down to two entities were blocked in 1998 by antitrust action. 8 A significant motive for the merger wave among drug manufacturers was the desire of companies to pool their research and development project portfolios, yielding cross-fertilization of specialized know-how and hoped-for risk hedging. On the former, see Henderson and Cockburn (1996). 4. Research, development, and the discovery of new drugs9 Research and the development of new and improved products play a central role in the activities of modem pharmaceutical enterprises. This was not always so. There have been several revolutions in the way new drug products are created. See, e.g., Schwartzman (1976, Chapter 2), Cooper (1970), and Temin (1979b). As late as the 1930s, the use of scientific methods to develop new medications was rare. Millennia of experience had identified naturally occurring substances with therapeutic properties. Co-existing with genuinely efficacious medicines were numerous 8 "Judge rejects two separate drug mergers", New York Times, August 1,1998, p. B1. 9 Substantial parts of this section are adapted from Scherer (1996, pp. 343-346).

1306

FM.Scherer

quack medicines. Retail pharmacies' shelves were lined with bottles containing organic and inorganic chemicals compounded on the spot to satisfy physicians' prescriptions or patients' pleas for the druggist's own preferred recipe. Until 1938, prescriptions were required in the United States only to obtain narcotic substances. The synthesis of aspirin marked one of the first breakthroughs in the use of modem chemistry for medicinal purposes. See Mann and Plummer (1991). A substance from the bark of the white willow tree had long been used to relieve fever and pain. By the 1830s German and French chemists had extracted its active ingredient and derived from it salicylic acid. But salicylic acid caused ulcers and other gastric problems. Seeking new markets for by-products of the synthetic organic dyes upon which its growth was based, the Bayer company of Germany established in 1896 a laboratory devoted to creating and testing dyestuff derivatives for medicinal effects. Early work led to the synthesis of acetylsalicylic acid, which it named "Aspirin" (a trademark that even a century later can be used in Germany only by Bayer) and sold with great success. In the early years of the 20th Century a German professor, Paul Ehrlich, formulated a conception of how small organic molecules interacted with proteins in the human body as keys do with locks. Ehrlich found many new chemicals with useful therapeutic effects, including Salvarsan, the first drug effective against previously incurable syphilis. Later research at the laboratory of Bayer's merged successor, I.G. Farben, led in 1935 to the discovery of a red dye derivative, sulfanilamide, that combatted lethal streptococcal infections. Numerous sulfa drug variants were subsequently synthesized and tested, leading to safer anti-infectives and the discovery of drugs with diuretic (blood pressurereducing) properties. The antibacterial properties of a naturally occurring mold, penicillium notatum, were first observed accidentally in 1929 by Alexander Fleming in England. Fleming failed to follow through, but penicillin's therapeutic properties were identified by Howard Florey and Emest Chain at Oxford University in time for that first antibiotic to play a lifesaving role in treating World War II casualties. Mass production methods using corn steep liquor as a fennentation medium were devised at a U.S. Department of Agriculture laboratory. Twenty U.S. companies participated in the top-priority wartime penicillin production program. When the war ended, intense price competition among those companies drove the wholesale price of penicillin in 300,000-unit doses from $3 in 1945 to $1 in 1948 and 10 cents in 1953. See U.S. Federal Trade Commission (1958). The therapeutic success of penicillin suggested to Selman Waksman of Rutgers University that other naturally occurring spores might have antibiotic effects. By screening and testing numerous soil samples during the early 1940s, he made two important discoveries: a specific new antibiotic, streptomycin, and even more important, a systematic method for finding new medicinal substances. Waksman insisted that the patent he obtained on his purified form of streptomycin be licensed to all interested parties, and as a result, the penicillin price competition experience was repeated, leaving the average producer with negative profits. Nevertheless, traditional pharmaceutical companies found in Waksman's methodology a potent means of discovering new therapeutic entities on which they might obtain

Ch. 25:


1307

exclusive patent rights. Several new broad-spectrum antibiotics - Aureomycin (chlortetracycline), Chloromycetin (chloramphenicol), oxytetracycline, and tetracycline - followed between 1948 and 1953. Each was patented, each was sold at prices high in relation to its production costs, and each proved to be very profitable for its suppliers. The lure of high profits in turn induced many other firms to try their hand at the game. U.S. pharmaceutical company R&D outlays (at constant 1990 price levels) rose from roughly $263 million in 1951 to $1.7 billion in 1967, $3.1 billion in 1980, and $8 billion in 1990. With few exceptions, the industry's approach to research and development during this period entailed testing for therapeutic effects large numbers of newly synthesized molecules (following the Ehrlich model) or naturally occurring substances (the Waksman model). In 1970 alone, U.S. pharmaceutical manufacturers are said to have conducted more than 700,000 laboratory culture and animal screening tests, out of which only a thousand chemical entities emerged with sufficiently interesting results to be carried into higher animal and human tests [Schwartzman (1976, p. 60)]. To the extent that this process was guided by systematic theory, it stemmed from prior discoveries that certain molecules had recognized therapeutic effects, so chemists synthesized "me too" variants of those molecules to see whether they might work better or at least circumvent rival firms' patents. As scientific knowledge has advanced, the industry has moved gradually from more or less random screening to methods called "rational drug design". Extending the early insights of Paul Ehrlich, it is known that each of the countless proteins in the human body has specific functions, and that the functioning or malfunctioning of those proteins is sensitive to the addition of chemicals at key receptor surfaces. The properties of receptors can be gleaned through various analytic techniques, and molecules can be designed to fit the receptors. Breakthroughs in genetic research and techniques for creating and multiplying new life forms have made it possible to manufacture proteins and other substances that replace defective proteins in the body or modify their functioning. Although much more highly focused than the "try every bottle on the shelf" methods of the 1950s and 1960s, these modern techniques are more like shotgun blasts than rifle shots. Considerable trial and error, including animal and then human tests, are required to isolate among many possibilities new molecules or proteins that actually work without serious adverse side effects. Failure is still more frequent than success. I The science base upon which pharmaceutical manufacturers build in discovering new drugs owes its growth in part to private research investments. Among the 24 U.S. industry groups on which detailed survey statistics are published, pharmaceuticals devoted the highest fraction - 16.6 percent - of its total 1992 R&D outlays to basic research [U.S. National Science Foundation (1996, p. 44)]. For all other industries, the comparable average was 5.3 percent. Even more important, however, is the role of government, e.g., in the United States, the National Institutes of Health (NIH), which perform

10 On the extent to which scientific research can reduce drug development uncertainties, compare Gambardella (1995) with Schwartzman (1997).

1308

F.M. Scherer

basic research and some applied work (including testing cancer drugs) intra-murally and award grants supporting much of the relevant research at universities and medical schools. In 1991, NIH research spending and grants totalled $7.7 billion, of which as much as $4.8 billion was in fields germane to pharmaceuticals [Scherer (1995, p. 27)]. This work was supplemented by $260 million of National Science Foundation grants in the biological sciences along with appreciable additional sums from other federal agencies. Knowledge flows from basic research institutions to companies not only through the medium of publications, but also through direct collaboration. Rich scientific paper co-authorship interactions between drug company research staff on one hand and academic and governmental basic researchers on the other have been traced by Cockburn and Henderson ( 19 9 8 ).l l Edwin Mansfield (1995) questioned a panel of U.S. industrial R&D decision-makers concerning the relevance of academic research to the development of specific new products. For the pharmaceutical industry, he learned that 27 percent of the new products sampled could not have been developed absent underlying academic research, and an additional 29 percent of the products were "significantly facilitated" by academic research. Academic research was found by Mansfield to be more important to the emergence of new products in pharmaceuticals than in the computer, instruments, electrical equipment, and metals industries. An econometric study by Toole (1997) suggests that 17 years elapsed on average between basic biomedical research and the introduction of commercial products building upon that research. See also (on oral contraceptives) IIT Research Institute (1968, pp. 58-73). Government agencies also provide important support for pharmaceutical research and development through educational grants for prospective scientists. In 1989, the U.S. National Institutes of Health had training grants of roughly $327 million outstanding, supporting some 7,800 pre-doctoral candidates and 6,600 post-doctoral researchers in the life sciences [U.S. Office of Technology Assessment (1993, p. 205)]. Many of the researchers trained in this way subsequently accept positions with pharmaceutical and biotechnology firms, carrying both the skills they have honed and the knowledge of technological opportunities they have acquired into commercial research and development activities.

5. Government regulation of new drug introductions Most of the industrialized nations have created regulatory institutions to ensure that desired safety and efficacy standards are met by new and existing drugs (and also by certain medical devices). The severity of regulatory control varies widely among national

11 Analyses of citations in published scientific articles to other work reveal that knowledge flows are international, but with appreciable own-nation biases. See U.S. National Science Board (1998, p. A-327).

Ch. 25:

1309


jurisdictions, with the United States leading chronologically in the implementation of tough standards. We focus therefore initially on the U.S. experience. 12 Systematic regulation began in the United States with the Pure Food and Drug Act of 1906, which prohibited adulteration, imposed labelling mandates, and required that drugs sold in interstate commerce conform to U.S. Pharmacopeia standards. After some hundred deaths occurred when a drug maker dissolved sulfanilamide in toxic diethylene glycol, the 1938 Food, Drug, and Cosmetic Act was passed, giving the Food and Drug Administration (the FDA, established in 1930) power to bar the interstate sale of new drugs unless they were found to be safe. However, if the FDA did not act upon applications for new drug approvals (NDAs) within 180 days, the drug's maker was authorized to begin commercial sales without regulatory permission. Despite the FDA's weak position, an FDA staff member successfully delayed in 1961 the transition of a new tranquilizer, thalidomide, from clinical tests into full-scale commercial marketing. It became clear shortly thereafter that sales of the drug in Europe, where thalidomide was approved and used widely to combat morning sickness, had led to the birth of some 8,000 malformed babies. Thalidomide became a cause celebre, leading the U.S. Congress to reorient an ongoing investigation of drug prices and profits toward legislation strengthening regulatory controls. The Kefauver-Harris Act of 1962 eliminated the 180-day loophole, required the FDA to certify that new drugs were not only safe, but also efficacious, and gave the FDA substantial new powers over drug testing and marketing. Under regulations subsequently articulated by the FDA, an organization seeking to test a new chemical entity in human beings must first obtain from the FDA an "Investigation of New Drug" (IND) authorization. IND applicants must submit evidence of non-toxicity at anticipated dosage levels from tests in animals such as dogs or monkeys and describe in detail their contemplated experimental design, including controls that will ensure statistical validity. The FDA has 30 days from the IND application's submission to raise objections. When clinical testing begins, it is typically arrayed over the following three-phase sequence, with median time spans and attrition rates (conditional upon passing through the prior phase) estimated from 93 clinical trials conducted between 1970 and 1982:13

Phase I: Administration to a small number of healthy volunteers to test for absorption, metabolism, and (at varying dosages) toxicity. Phase n: Administration to a few and then dozens of patients with the disease to be treated. Phase III: The drug is administered in double-blind tests to at least two large samples of patients with the disease. Long-term toxicity tests are conducted in parallel.

Months (median)

Attrition rate

15.5

25%

24.3

52%

36.0

36%

12 The next several paragraphs are drawn in part from Scherer (1996, pp. 346-350). 13 The source is DiMasi et al. (1991). The times and attrition rates ignore possible overlaps between stages.

1310

EM. Scherer

If the drug clears all of these hurdles successfully, its sponsor applies to the FDA for a new drug approval (NDA). The FDA often requests additional tests and information before making its decision. On average, for the 93 clinical trials summarized above, 23 percent of the chemical entities that entered Phase I testing emerged with FDA approval. The median lag for that sample from NDA application (following Phase III) to FDA approval or disapproval was 30 months. Adjusting for overlapping phases, the median time span from the start of clinical testing to NDA issuance was 98.9 months, or slightly more than eight years. The total amount of time required for successful drugs to pass through these testing and regulatory hurdles has tended to increase in successive decades since the Kefauver-Harris Act was passed - e.g., from 4.7 years on average in the 1960s, 6.7 years in the 1970s, 8.5 years in the 1980s, and 9.1 years in the mid-1990s [DiMasi et al. (1994) and Kaitin (1997)]. The increase in clinical testing mandates and the time needed to satisfy them led to striking increases in the average cost of clinical trials. A study by Mansfield (1970) of 17 new drug development projects before the 1962 law took effect found an average success rate of 37 percent. The average development and testing cost per successful NDA, including the pro-rated costs of failed chemical entities, was $1.05 million (in current dollars, equivalent to roughly $4 million in 1987 dollars). For the 93 projects studied by DiMasi et al. (1991) whose histories are summarized above, the average clinical development cost per NDA had escalated to $48 million (in 1987 dollars).' 4 Correcting for general price inflation, the increase in testing costs between the late 1950s and the late 1970s was roughly ten-fold. Grabowski et al. (1978) took advantage of a difference between U.S. and British testing regulations to illuminate the reasons for increasing costs. U.K. regulations were strengthened in 1971 from reviewing safety only, as had been FDA practice up to 1962, to requiring proof of efficacy also. Between 1960-1961 and 1966-1970, average inflation-adjusted drug development costs in Great Britain rose by a factor of three, while those in the United States increased sixfold. This suggested that more stringent regulation in the United States was responsible for a twofold increase, while influences common to both regulatory jurisdictions explained the remaining threefold increase. Among those other influences were the recognition by drug companies that extensive testing was required to avoid repeating the thalidomide disaster, which imposed huge tort liability losses on European firms; the need to accumulate evidence used in persuading physicians to favor new drugs over tried-and-true older formulations; and perhaps also increasing difficulty in finding superior new drugs following the first wave of successes experienced during the 1940s and 1950s.

14 This estimate is for out-of-pocket clinical testing period costs only. Costs incurred in the pre-clinical discovery phases, when pro-rated over successful NDAs, approximately double the total R&D cost of a marketed NDA. It has become customary in the literature to take into account the opportunity cost of invested funds by capitalizing phased R&D costs to the time of market approval, which raises even more the relative weight of pre-clinical work, farthest in time from the date of a new product's introduction. Capitalized R&D costs as high as $200 million per successful NCE are commonly reported.

Ch. 25:


1311

i

/U

I

I

..... ... .......2....2....!... ......... :

0 60 QA) cL

w

50

E m

40

i

I

I--- I .

.......

.......+

_ .........

All NCEs . .... Important

............. .. ........

0 (a

Ju

o

C

20

E Z

10n .

n

I

1940

.

..

I

1950

*.

C.

-l

1960

·

,

1970

1980

1990

Year Figure 1.Trends in U.S. new drug approvals, 1940-1990.

Despite rapidly rising aggregate drug R&D expenditures, the number of new drugs approved for marketing in the United States declined markedly. This is shown by Figure 1.15 The solid line, counting the number of new chemical entity approvals, reveals NDAs rising erratically and then peaking at 65 in 1959, only to plummet and hover around an average of 18 per year following enactment of the Kefauver-Harris law in 1962. The Food and Drug Administration was criticized for the drop in new drug appearances and later for "drug lag" - the slow approval of even those drugs that ultimately did reach the market, causing the United States to fall behind other nations in making the most up-to-date therapies available to its consumers. See, e.g., Peltzman (1973), Wardell and Lasagna (1975), and U.S. House of Representatives (1980). Peltzman laid out inter alia a theoretical schema for FDA regulation, whose principal benefit, he argued, is to eliminate what might otherwise be information failures in the marketplace. It is illustrated in Figure 2. Suppose the demand curve reflecting some combination of patient and prescribing physician preferences without testing information required by the FDA is BDUNc. With no FDA intervention, consumers will consume OXu units of the drug per year at a price (assumed fixed over all cases) of OP. If, however, the true fullinformation demand curve DT1 lies much more to the left because FDA-mandated testing reveals adverse side effects, consumers with full information would consume only 15 It is drawn from Scherer (1996, p. 351), tapping a variety of sources.

1312

EM. Scherer

0)

au 0

0

CL

IlUlIIULI

Ul

rIl :

.sll4LlUl4

l

Figure 2. Welfare consequences of uncertainly reduction through clinical testing.

OXI units. Over-consumption due to imperfect information causes a consumers' surplus loss measured by near triangle GFDT1, offset to a modest extent by the sacrifice of consumers' surplus EGP during the period when the drug is not available to consumers because of FDA-mandated testing. If on the other hand FDA-mandated tests show the drug to be better than poorly-informed decision-makers' beliefs, the full-information demand curve will shift to ADT2 and consumption will increase to OX 2, adding consumers' surplus triangle KCF. However, if the tests are so costly that companies choose not to develop the new drug, consumers will lose trapezoidal surplus AKFP as a result of FDA intervention. From a statistical analysis of the decline in drug approvals, and making the key assumption that the decrease evident as early as 1960 would in time have been reversed, as earlier declines were, Peltzman concludes that the 1962 drug law caused a substantial loss of consumer benefits. In response to its critics, the FDA insisted that at least part of the sharp drop in NDAs was intentional. What it had done by requiring more rigorous and costly testing, its officials said, was mainly to discourage "me too" variants adding little or nothing to the therapeutic benefits of already existing drugs. It supported this argument by pointing to its internal ratings of newly approved drugs according to their therapeutic novelty. The dotted line in Figure 1 tracks the appearance of drugs considered by the FDA to offer important therapeutic gains. In the 1950s, there was a surge of important drug discoveries as the drug-finding revolution took hold, but then, long before 1962, the appearance of important new chemical entities stabilized in the range of zero to six NDAs

Ch. 25:

1313

The PhannaceuticalIndustry 4

n.

I UV

..

....

...

- - ...

.

-

.

...

-

-

.

.........................

--------I ---,

- ---. ........

· .I

......

.......

...

--

... '' - - -

I

v#

a0

cn Ca

90 0

oo

- -.

80

-.

1111...

I -

- - - ...

70 0

2)

60

a0

50

...........

....

-

L

Ca

40

13,

30 (n ........-

x~3 3 uJ, a

.....

10n~2

USA

Switzerland Germany UK

France Italy

Sweden Japan Netherlands

Nation Figure 3. Percent of multinational drug company sales outside home market. Source: Thomas (1996).

per year, with a mean of 3.6. What had been weeded out, the FDA continued, were mainly the drugs yielding little or no therapeutic gain. Critics questioned the accuracy of the FDA's drug importance evaluations. Industry spokesmen acknowledged that they had refocused their R&D efforts away from "me-too" drugs and toward the goal, not always achieved, of developing only therapeutically important molecules. A retrospective analysis by L.G. Thomas (1996) injects surprising new insights into this debate. For a while, the United States had the most stringent drug testing requirements in the world. The United Kingdom and Germany were the next major nations to follow the U.S. lead. By setting high standards for new drugs, Thomas argues, the regulatory agencies in those nations forced their domestic drug firms to target their R&D on drugs of superior therapeutic efficacy, and when success was attained, the resulting drugs sold well not only at home but also in foreign markets. The data Thomas compiled in support of his argument are summarized in Figure 3, which shows the fraction of the 1985 sales of companies at home in nine leading drug-developing nations realized outside the firms' home markets. The smaller a nation's home market, the larger was the fraction of sales realized away from home. This plus the fact that all three nations have excellent university systems explains the relatively high external sales ratios for Switzerland, Sweden, and the Netherlands. Taking into account this population effect, the stronger the home regulatory regime, the larger companies' sales were outside their home markets. France, Italy, and Japan, with populations as large as those of (West) Germany and the United Kingdom, had unusually weak domestic testing standards, which

1314

FEM. Scherer

are reflected in poor external market performance. The United States, Germany, and the United Kingdom, with tough standards, all do well. 1 6 Despite this inadvertent mercantilist policy success, the U.S. Congress continued to be alarmed over "drug lag" claims and especially the long decision-making lags experienced when applications for a New Drug Approval reached the FDA. A new law passed in 1992 and amended in 1997 permitted the FDA to collect fees on new drug approval filings and also annually from drug-producing plants and from drugs approved after being in the FDA's review "pipeline" by September 1992 or later. See Shulman and Kaitin (1996). The proceeds of these fees, totalling roughly $325 million between 1993 and 1997, were used to augment the FDA's technical staff, as a consequence of which the FDA agreed to a time-phased program of reaching an increasing fraction of its decisions within one year on conventional NDA applications and even more quickly for high-priority drugs. An appreciable reduction in decision-making intervals in fact resulted, from an average of 31 months over 1987-1991 to 16 months in 1997 [PhRMA (1998, p. 24)]. It and several highly publicized withdrawals of approved products from the market precipitated criticism that FDA approval standards may have been reduced in the bargain. The acceleration of decision-making and clearing out of backlogs also led to 53 new drug approvals in 1996, compared to an average of 23 in 1987-1991. Regulatory reforms also occurred in other industrialized nations. The most important change was the creation, beginning in January 1995, of the European Medicinal Products Evaluation Agency (EMEA), which provided a uniform forum for regulating new drug introductions in all member nations of the European common market. See Heppell (1996). Initially, only medicines developed using new biotechnology methods were placed within the sole jurisdiction of EMEA. For other products, manufacturers retained the option of seeking approval either before EMEA or with national regulatory authorities, whose decisions would be mutually recognized in all European Union nations. It is unclear whether this procedure led to convergence of national authorities' standards - e.g., whether greater stringency was introduced into what critics called the "French impressionist school of safety regulation" [Thomas (1996, p. 116)]. Beginning in January 1998, the mutual recognition approach was to be phased out and manufacturers seeking to market new products in more than one European nation were required to have their test results reviewed by EMEA. There has been too little experience with EMEA's decisions to ascertain whether they differ substantively from those of the more stringent national jurisdictions. 17 Differences between regulatory regimes point to several more fundamental philosophical questions. Granting or denying permission for full-scale marketing of a new drug is an exercise in decision-making under uncertainty. Clinical test insights can

16 See also PhRMA (1998, p. 71). 17 In a 1995 seminar at Harvard University, an EMEA official said that his agency was likely to slant its tradeoffs somewhat differently from those of the U.S. FDA, placing relatively more emphasis on speed of new product availability and relatively less on conclusive proof of efficacy and safety.

Ch. 25:


1315

be sharpened through the use of appropriate statistical methodologies, but they cannot eliminate uncertainty, especially for adverse side effects of very low incidence. 18 Decision-makers must weigh the risk of Type I errors - approving a drug when it is not truly safe or effective - against the risk of Type II errors - withholding from the market entities that are truly effective and safe. See, e.g., Scherer (1996, pp. 353-355). In governmental agencies, there is a natural tendency toward placing more weight on avoiding Type I errors, since officials who have approved a product that leads to cancer or malformed babies will be singled out for castigation in public fora. The tradeoff can be narrowed by increasing the size of clinical trial samples, but that solution increases costs, possibly discouraging the development of some drugs, and it is likely also to delay the availability of new drugs. Several further questions follow. If uncertainty is high but the possibility of lifesaving benefits is also substantial, shouldn't a regulatory agency illuminate the problematic tradeoff and let individual physicians and/or patients make their own risk-taking decisions, rather than being restrained by the choice of a bureaucracy? Stung by criticism that its decision-making was denying potentially vital therapies to patients with life-threatening diseases, the U.S. FDA began during the 1980s to make increasing use of "compassionate NDA" procedures under which experimental drugs that have not yet been approved formally are made available to physicians. For HIV drugs, it also waived the requirement that double-blind tests be conducted, for to assign a patient randomly to the placebo group could be tantamount to imposing a death sentence. Recognizing such exceptions, one must ask the further question, why should a regulatory agency be the ultimate decision-maker on whether any new drug can be used? To be sure, absent regulatory requirements, drug manufacturers might perform too little clinical testing to ascertain whether a drug is superior to existing alternatives. Meager testing was the norm in the pre-thalidomide era. An information market failure may need correction. But why doesn't the regulator merely require appropriate testing and disclosure of test data, letting physicians decide from the data whether the drug is safe and efficacious? If there is an argument for regulation of whether new drugs may be marketed, it must lie in a further information market failure - e.g., from the possibility that most physicians are too busy to make well-informed independent decisions. Carrying the debate one step farther, why should prescriptionsbe required to obtain drugs? They are not required for over-the-counter drugs or, in many less-developed nations, for any available drug. The prescription system implies that patients are unable to make well-informed decisions about their own welfare, so physicians must act in loco parentis. That may be true, but obtaining a prescription imposes costs, and Sam Peltzman's (1987) statistical analysis suggests that there is no clear indication of higher

18 Withdrawal of approved drugs because of safety problems is fairly rare. Three to four percent of the drugs approved for marketing between 1974 and 1993 in the United States, the United Kingdom, and Spain were subsequently withdrawn for safety reasons. The number of withdrawals was 10 in the United States, 16 in Spain, and 20 in the U.K. [Bakke et al. (1995)].

1316

EM. Scherer

poisoning or mortality rates in middle-income nations where prescriptions are not required. Similar perplexities are encountered in the regulation of medical devices. We focus only on the U.S. experience. Reacting to numerous deaths and maladies attributed to faulty artificial heart valves, cardiac pacemakers, and intrauterine devices, the U.S. Congress in 1976 increased the FDA's regulatory authority over the marketing of medical devices and therapeutic apparatus. See U.S. Office of Technology Assessment (1984, Chapter 5). All such devices were to be classified by the FDA into three categories, depending upon safety and effectiveness risks. Devices already on the market were reviewed by FDA panels, and those assigned to the highest-risk Class III were required to submit safety and efficacy data to the FDA, with forced cessation of marketing resulting for devices found to be unsafe. Newly-developed Class III devices must run a gamut of tests similar to those for new drug chemical entities before receiving marketing approval from the FDA. Lower-risk class devices are subject to manufacturing and performance standards published by the FDA. As in the history of drug regulation, there were complaints that FDA test and approval procedures impose excessive costs, discourage innovation, and delay the availability of potentially beneficial devices. In 1997 an unsuccessful attempt was made in the U.S. Congress to privatize significant elements of the FDA's device certification functions. 6. Patents and pharmaceutical innovation Owing to a combination of regulatory mandates and intrinsic difficulty whose exact weights remain unknown, developing new drug products is a costly, high-stakes game. It is also a risky game. The risks are shown in part through evidence assembled by Grabowski and Vernon (1990, 1994), summarized here in Figure 4. They compiled data on the domestic and foreign sales of 100 new drug chemical entities (NCEs) developed by U.S. companies and introduced into the U.S. market (following FDA approval) during the 1970s. Using standard ratios, they subtracted estimated production, distribution, and marketing costs and plant investments from product sales, leaving estimates of the "contribution margin" or Marshallian quasi-rents left to repay research and development investments and provide a profit. The 100 products' discounted quasi-rents (assuming 1986 price levels) were arrayed into deciles in descending order of absolute magnitude. What emerges is a highly skew distribution shown by Scherer (1998) to approximate log normality and to resemble the payoff distributions in other high-technology fields. The most lucrative ten NCEs contributed 55 percent of total quasi-rents. Products in seven of the ten deciles did not even repay their capitalized research and development investments (including the pro-rated cost of projects yielding no marketable product), whose average value was found to be $81 million per NCE.19 Products in the third decile came 19 A positive but statistically insignificant correlation was found between the magnitude of quasi-rents and individual product R&D costs.

Ch. 25: ) n


1317

-)UU

400 0 z 0

300

r e)

200

-

100

0)

o 0 UW Y

A1 v

Average R&D Cost _ ................ I.................................................

F-

1

2

3

4

5

6

7

8

9

10

Decile Figure 4. Distribution of 1970s drug NCE quasi-rents. Source: Grabowski and Vernon (1990).

close to break-even, while those in the second decile provided returns roughly twice average R&D investments. Summed over 100 new products, the discounted present value of quasi-rents exceeded the sum of R&D costs by only three percent. Thus, a relatively few "winners", and especially the "blockbusters" in the first decile, had to repay the losses on the majority of low-payoff products. 6.1. The unusual importance of drugpatents Although altruistic motives undoubtedly enter, profit is the principal lure leading drug makers to invest large sums toward the discovery and development of new drugs. Specifically, company leaders hope to develop products sufficiently important, and sufficiently well insulated from competition, to repay or more than repay their R&D investments. Patent protection on marketed new chemical entities is a significant component of their profit-earning expectations. Indeed, surveys asking research and development managers what factors permit them to reap the profit benefits from their innovations - including the competitive advantages of being first on the market with a new product, superior sales and service efforts, secrecy and complexity of product and production process technology, and patents - consistently show the pharmaceuticals industry to be one in which the greatest stress is placed on patent protection [Levin et al. (1987); Cohen et al.

1318

EM. Scherer

(1997)].20 In most industries, imitators must duplicate much or all of the R&D to field a product competitive with the original innovation, and this plus a head start and the ability to build brand loyalty in the interim are often sufficient to make R&D investments profitable. 21 But in pharmaceuticals, much of the R&D investment is incurred to discover which molecules have medically interesting properties and to test their efficacy in humans. Absent patent protection or some equivalent barrier,22 imitators could free-ride on the information created by the innovator's hundred-million-dollar R&D and testing investment, spend a few hundred thousand dollars on process engineering, and begin competing with the innovator, eroding its quasi-rents. 2 3 At the same time, drug product patents, unlike the patents in many other fields of technology, protect a clearly identified chemical molecule around which the marketing of substitute variants is impossible without undergoing a complete new array of FDA clinical trials. The development of significant new drugs is concentrated largely in the nine highly industrialized nations covered by Figure 3 [Ballance et al. (1992, pp. 86-88)]. Many other nations, and especially the less-developed countries, have traditionally denied patent protection to medical and (frequently) food products. In this way, they are able to avoid the monopoly prices and drain of scarce foreign exchange that would occur if patent protection were granted and the patented drugs were imported (1992). Even Switzerland, home to three of the world's leading pharmaceutical companies, provided no drug product patent protection until 1977. A few nations allowing no drug product patents - notably, during the 1960s and 1970s, Italy, and more recently, India - evolved domestic pharmaceutical industries adept at "knocking off" new drugs still patented in industrialized nations and selling them at bargain prices both at home and (through exports) in other patent-denying nations. See, e.g., Lanjouw (1997). 6.2. Consequences of the Uruguay Round agreement Even though the half of the world's population living in underdeveloped nations account for less than 20 percent of world pharmaceutical consumption, the patent policies of LDCs were viewed as a thorn in the sides of multinational pharmaceutical producers. During the 1980s the United States government, urged on by a lobby organized by pharmaceutical executives, began threatening such nations with international trade 20 The only industry in the second [Cohen et al. (1997)] survey placing more stress on patents than the pharmaceutical industry was medical equipment. 21 On brand loyalty as a barrier to imitation of innovative drug products, see Bond and Lean (1977). 22 On new drug approval as a regulatory barrier to imitation see Kitch (1973). In the Orphan Drug Act of 1983, the U.S. Congress ordained a seven-year period of exclusivity following FDA approval of drugs treating symptoms affecting fewer than 200,000 persons, whether or not patent protection co-exists. This orphan drug privilege serves as a surrogate patent. Between 1983 and 1998, 140 drugs were approved by the FDA under orphan drug rules. 23 There are exceptions. Some drugs are very difficult to produce. An example is the cephalosporin antibiotic cefaclor, whose production entails ten complex steps and on which competitive imitation was retarded even after a product patent and several key process patents expired during the early 1990s.

Ch. 25:


1319

sanctions under Section 301 of the U.S. trade law unless they conformed to U.S. intellectual property standards. See Santoro (1992) and Ryan (1998). Several nations changed their laws to comply, and to pave the way for a bilateral trade treaty, Canada amended its laws in 1987 to end compulsory licensing of drug product patents at quite modest royalties, which had led to relatively low new drug product prices in Canada. See McFetridge (1997). U.S. pharmaceutical industry leaders extended their lobbying campaign to include compatriots in Europe and Japan and also attracted the collaboration of motion picture, musical recording, and software interests. Their governments responded by making harmonization of intellectual property standards a high-priority item on the agenda of the Uruguay Round trade negotiations. The effort was successful. The resulting Treaty of Marrakech, ratified in April 1994, requires inter alia that all signatory nations provide full patent protection for pharmaceutical products - for industrialized nations, by 1999, and for less-developed nations, by the year 2004. These measures continued to be controversial even after the Marrakech Treaty was signed. The parliaments of some LDCs balked at enacting the required law changes. For their nations, the benefits from a modest increase in the number of new drugs likely to be forthcoming as a result of broader international patent coverage were plausibly believed to be outweighed by the loss of consumers' and producers' surpluses from having to import at elevated prices the most up-to-date drugs from multinational companies. See Deardorff (1992). The balance might change if granting drug product patents led to the emergence of a local pharmaceutical industry skilled at discovering and developing innovative new drugs for local production and export. However, the experience of Italy, whose Supreme Court ordered in 1978 that the government begin issuing drug product patents, undermining a foundation of Italy's strong "knock-off" industry, suggests that making the transition from imitative to innovative industry is at least time-consuming and perhaps even unlikely [Scherer and Weisburst (1995)]. Canada's experience after it strengthened drug patent rights was more favorable because it was able to extract from the multinational companies a pledge to move a substantial fraction of their R&D activity to Canada and because Canada had a surplus of well-trained life scientists upon which that shift could build. See McFetridge (1997).

7. Pricing Although companies selling new drug chemical entities commonly enjoy patent protection for a number of years after their product is introduced, more often than not they must face competition from chemically differentiated molecules that might be prescribed to treat the same symptoms. The pharmacy benefit management company PAID published a formulary for 1993 that can be subdivided, applying some judgment, into 141 specific symptom groups. 2 4 The number of drugs per symptom group ranged from 24 PAID Prescriptions Inc. (1993), The PAID National Formulary (Montrale, NJ). Some sub-categories were combined where it was clear that the same symptoms were being treated in the same general mode.

1320

EM. Scherer

one to 50, with a median of five drugs and mean of 6.04. Thus, the typical market structure for first-line drugs is differentiated oligopoly. How are prices set under those circumstances? 7.1. Pricingbranded drugs The most thorough relevant study is by Lu and Comanor (1998), who analyzed the pricing of 148 new branded chemical entities introduced into the U.S. market between 1978 and 1987.25 All but 13 had at least one fairly close substitute in their principal therapeutic indications. The average number of substitutes was 1.86. The authors distinguished among other things between drugs that offered an important therapeutic gain, as evaluated by the FDA (10 percent of the sample); those providing modest gains (37 percent); and those offering little or no gain relative to substances already on the market. Lu and Comanor found that drugs contributing important therapeutic gains were introduced at prices 3.2 times the level of substitute products in equivalent dosages; those offering modest gains were priced at 2.17 times the average for substitutes; those making little or no gain were priced at roughly the same level as pre-existing substitutes. During the four years after introduction, prices of the important new drugs (adjusted for general inflation) tended to decline by about 13 percent on average, while the prices of drugs making little or no therapeutic contribution rose on average by 22 percent. Introductory prices tended to be lower by 8 to 10 percent, all else equal, for each additional competing differentiated substitute available at the time of initial marketing. 7.2. Tort liability risks andprices Prices are also influenced by the amount of tort liability risk borne by drug manufacturers. Such risks vary both with drugs' inherent chemical properties and with national or state legal rules. Controlling for these two variables with a matched sample of drugs sold both in Canada and the United States, Manning (1997) found that roughly half the difference between the prices of identical drugs in those two nations was attributable to drug-specific tort liability risk, which was generally higher under U.S. legal precedents. Because they inject modified actual disease vectors into the human body, vaccines pose particularly high risks of unfavorable outcomes 2 6 - so much so that public health authorities in the United States experienced difficulty during the early 1980s in obtaining supplies for immunization programs. See U.S. Office of Technology Assessment (1993, pp. 176-182), Garber (1993), and Manning (1994). As a consequence, the U.S. Congress passed in 1986 the National Childhood Vaccine Injury Act, which reduced companies' risk exposure by adopting a no-fault compensation scheme designed 25 See also Schwartzman (1976, Chapter 12). This paragraph is drawn with minor changes from Scherer (1996, p. 369). 26 The author's bunk mate died, and the author came close to death, as a result of defective vaccines administered at a U.S. Army basic training camp in 1954.

Ch. 25:


1321

to minimize the frequency of large damages awards based upon traditional negligence theories. 7.3. Generic drug competition Once patents expire, a new form of competition may emerge - competition from generic drugs, that is, products with the same active chemical ingredient(s) as the original pioneer drug, and which are normally sold with little or no advertising or field sales promotion. The extent to which generic drugs are substituted for original branded drugs and their impact on prices varies widely from nation to nation and also across therapeutic categories. A survey by Ballance et al. (1992, p. 210) reports little use of generic drugs in Belgium, France, and Spain during the early 1990s, but extensive and rising use in the United States, Denmark, Germany, and the United Kingdom. See also Perry (1996). We focus initially on the U.S. experience. As the post-World War II drug revolution gained force, most states in the United States had laws limiting the ability of pharmacists to dispense anything other than the specific brand prescribed by a physician. Since physicians typically prescribed by brand name rather than by (less memorable) chemical name, this precluded the substitution of a generic for a prescribed branded drug, and generic use was modest. During the 1970s and early 1980s all states with such anti-substitution laws repealed them and passed new laws permitting (with wide variation from state to state) some generic substitution. But significant barriers to substitution remained. As of 1980, generics were dispensed on only one fourth of the prescriptions for which substitution was feasible [Masson and Steiner (1985, p. 26)]. There were two main limitations on generic use. For one, remaining legal hurdles, including obstacles as trivial as the design of the physician's Rx form, interacted with weak patient or pharmacist incentives as impediments to substitution [Masson and Steiner (1985, Chapters 4 and 5)]. These will be explored in a different context shortly. Second, FDA and judicial interpretations of the 1962 Kefauver-Harris Act required would-be generic drug producers seeking approval from the FDA after an innovative drug's patent expired to carry out tests nearly as extensive and costly as those associated with a new chemical entity. Since the generic products would sell at much lower prices than the original branded drug and might secure only modest market shares, generic suppliers were often deterred from undertaking the required effort. A grand compromise embodied in the Waxman-Hatch Act of 1984 remedied this problem along with one troubling the research-oriented drug manufacturers. Under the new law, generic suppliers were obliged only to demonstrate before the FDA that their drug had the same active ingredient(s), that their formulation was absorbed into the blood stream at a rate within plus-or-minus 20 percent of the original drug's norm (usually shown through tests on 24 human subjects), and that they adhered to good manufacturing practices. Moreover, they were allowed to manufacture test samples before the original drug's patents expired so that tests could be conducted and documentation submitted to the FDA in advance of patent expiration. Thus, they could hit the

1322

EM. Scherer

ground running - perhaps even on the day of patent expiration. As a quid pro quo for the branded drug makers, Congress authorized an extension of the patent protection period to compensate for the delays caused by FDA-required clinical testing regulations. 2 7 Typically, patents were obtained on promising new molecules at about the time when clinical testing began, so if the tests and FDA decision-making took eight years, only 17 - 8 = 9 years of patent life remained. 2 8 Under Waxman-Hatch, the period of exclusive protection could be extended by as much as five years, e.g., in the illustration here, to 14 years. 29 Thus, drug developers would be given a longer period of exclusive sales, but would have to face tougher competition once the period of exclusivity ended. Both features strengthened incentives for vigorous new drug development. 7.4. Branded vs. generic drug price competition The new legal mandates spurred a wave of entry into generic drug manufacturing, saving U.S. consumers (or their insurers) an estimated $8 to 10 billion in 1994 [U.S. Congressional Budget Office (1998, p. 31)]. Generics' share (by countable units, e.g., tablets) of U.S. prescription drug sales rose from 18.6 percent in 1984 to 32.9 percent in 1990 and 44.3 percent in 1998 [PhRMA (1998, p. 57)]. Their share of dollar sales at retail was much lower, e.g., 17.3 percent (vs. 36 percent by number of prescriptions) in 1994,3 ° in part because the newest and highest-priced drugs were still covered by patent protection and because, when generic entry began, the price regime for the typical drug bifurcated. Generics tended to enter the market at whole sale prices 40 to 70 percent of those prevailing before the original drug's patent expired. As additional generic competitors entered a product category, the generic price fell, e.g., to 29 percent of the pre-competition price with 10 generic rivals and 17 percent with 20 rivals [Caves et. al (1991, p. 118)]. Meanwhile, prices of the original branded drug remained essentially stable, according to the analysis of Caves et al. (1991) or even rose, according to Grabowski and Vernon (1992) and Frank and Salkever (1997). Figure 5 illustrates the not atypical history of pricing after the product patent covering the cephalosporin antibiotic cephalexin (sold as Keflex by the original patent holder, Eli Lilly) expired in April 1987. 3 1 Two price trajectories for generic entrants are shown, one averaging the wholesale prices charged by R&D-oriented firms (dotted line) for 100 250-mg capsules and one (dashed line) the comparable prices of generic specialists.

27 Japan enacted a similar law effective in January 1988, and in June 1992, European Community regulation 1768/92 authorized "supplementary protection certificates" with similar effect. See Ager (1997). 28 At the time, the normal life of a U.S. invention patent was 17 years from the time the patent was issued. As a consequence of the Marrakech Treaty, the law was changed in 1994 to make the patent life 20 years from the time an initial application is filed. The average time from application to issue was two to three years, but with considerable variation about the mean. 29 The average actual extension was 2.3 years [Tufts University (1997)]. 30 See U.S. Congressional Budget Office (1998, p. 15). 31 It is drawn from Griliches and Cockburn (1993, Figure 1).

Ch. 25:


1323

^^

YU

80 Branded Keflex o,

70

70 0

.,.,.,.,,,,,.

-- /,.................... -7 ".............................

..........

60 ..........

.............

.........

..........................

Generics ......... . .

o

1o50 n

I

19 85

i

I

86

I

I

87

88

q

i

I

89

i

I

90

i

I

91

Year Figure 5. Trends in cephalexin prices with generic entry.

Initial generic entry occurred at prices slightly less than half the branded product's price. As additional firms began generic sales (ultimately, more than 20), the average generic price was competed down to as little as 15 percent of the (rising) branded product price. The reason for this bifurcated pricing behavior is that some physicians retained strong preferences for the original brand and chose not to permit substitution. Branded drug manufacturers in effect confronted two markets - one consisting of price-insensitive consumers willing to pay high prices for the security of a brand name, and another (growing in relative size as the 1980s gave way to the 1990s) consisting of pricesensitive consumers willing to shift to generics. See Frank and Salkever (1992) and Scherer (1996, pp. 376-378). The branded drug suppliers found it more profitable to serve a minority fraction of their molecule's market (by 1992, 28 percent on average of total unit sales 3 2) at high prices than to reduce their prices to the low levels required to match generic competition. Some branded drug manufacturers practiced price discrimination in an attempt to serve both market segments. Price-insensitive consumers were sold the original brand, but to capture price-sensitive consumers, the patent holder introduced shortly before patent expiration a "branded generic" - i.e., the same drug under a different label, priced at lower levels than the original brand but higher than

32 See Grabowski and Vernon (1996, p. 114).

1324

EM. Scherer

no-name generics. In this way they gained a "first mover advantage" in the generic market, 33 secured the leading share of generic sales, and perhaps thereby discouraged some would-be generic suppliers from entering and driving prices even lower. However, most companies feared adopting such a strategy because it could accelerate the shift from their high-price branded products to low-priced generics. According to a U.S. Congressional Budget Office study (1998, p. 34), among 112 drugs with generic competition, the original brand holder sold its own generic product and gained a retail market share of more than 10 percent in only 13 cases. Not all of the savings from generic drug price reductions at wholesale are passed on to consumers in the form of lower retail prices. On average, retail pharmacies retain higher dollar margins, and hence much higher percentage margins, on the generic drugs they dispense than on the equivalent, higher-priced, branded drugs. See Masson and Steiner (1985, p. 36) and Grabowski and Vernon (1996, p. 117). Higher dollar margins are realized on generics in part to defray inventory holding costs. A more important reason, however, is that pharmacists must stock the leading branded product, but have some choice in determining which of several possible generics to stock and dispense. This appears to be a special case of a more general phenomenon described as the Steiner effect: retail margins tend to be lower on products with strong consumer pull, e.g., those that are heavily advertised, than on less-advertised items. See Steiner (1993). Kopp and Sheffet (1997) report a substantial relative decline in retailers' margins for drugs receiving direct-to-consumer advertising support between 1988 and 1991. Viagra, the drug product with perhaps the strongest early consumer awareness in history, was merchandised by retail drug discounters at near-zero margins as a means of attracting patronage more generally. 3 4 7.5. Stimulating generic substitution The extent of generic substitution depends upon several variables in addition to the number and identity of generic suppliers. Generics have been particularly unsuccessful in replacing branded drugs for which attaining precise diffusion rates within the body is crucial to effective therapy, e.g., in heart rhythm regulation and the administration of anticoagulants after a heart attack. Passing over such relatively rare cases, generic substitution was encouraged during the 1980s and 1990s in the United States by three phenomena: rapidly rising affiliation of physicians with cost-conscious health care organizations enforcing strong pro-generic policies; "maximum allowable cost" (MAC) reimbursement rules under which state Medicaid authorities reimbursed only the cost

33 On first-mover advantages more generally, see Bond and Lean (1977) and Robinson et al. (1994). The race to be a first mover led some generic producers to falsify test results and bribe FDA employees. See Morton (1997b). 34 "Kmart, Wal-Mart Compete with Teeny Viagra Prices", Reuters dispatch (obtained on Yahoo Finance), May 21, 1998.

Ch. 25:

The PharnaceuticalIndustry

1325

of the lowest-cost qualified generic substitute for a drug;3 5 and lower co-payments under private health insurance plans when patients accepted generic as compared to branded drugs. The experience of Canada sheds further light on variables affecting generic substitution. See Gorecki (1986) and McRae and Tapon (1985). Generics were available early in branded drug life cycles during the 1970s and 1980s as a result of compulsory drug patent licensing policies. The various Canadian provinces had widely differing rules with respect to generic substitution, and substitution rates varied as a result in identifiable ways. Quebec, New Brunswick, and Nova Scotia had especially low substitution rates because they reimbursed the full cost of any drugs dispensed under Canada's lower-income and above-65 health insurance programs, whereas provinces with high substitution rates reimbursed only the cost of the least expensive generic. Designation of specific generic drugs as interchangeable with branded counterparts on provincial formularies markedly increased substitution rates. In Quebec, unlike other provinces, pharmacists were not absolved of malpractice liability in dispensing formulary generics. This, like Quebec's requirement that pharmacists acquire the patient's consent before substituting, reduced generic drug usage. 7.6. New institutions;New power relationships During the 1980s and 1990s, new forces emerged in the United States to countervail and weaken manufacturers' power to maintain high prices on patented drugs. Crucial to these changes was the rapid growth of health maintenance organizations (HMOs) and pharmacy benefit management firms (PBMs). Glied (2000) shows in more detail, health maintenance organizations affiliate substantial numbers of physicians to provide comprehensive health care on a prepaid insurance basis. Patient enrollment in HMOs increased from 9.1 million in 1980 to 33.6 million in 1990 and 46 million in 1995 [U. S. Bureau of the Census (1996, p. 121)]. An even newer phenomenon was the emergence of PBMs to manage the payment paperwork on out-patient prescription drug insurance claims for HMOs, companies with large employee health care insurance plans, and conventional insurance carriers. By 1993, the six largest PBMs were managing an estimated 36 percent of all U.S. retail prescriptions [U.S. Congressional Budget Office (1996, p. 21)1. To control the rising costs of prescription drugs, many HMOs and traditional hospitals began establishing formularies listing the drugs suitable for use against particular illnesses. When appropriate generic drugs existed, formularies strongly encouraged affiliated HMO or hospital staff to use them in place of higher-priced branded drugs. But even when no generic substitutes were available, formulary committees began selecting from the menu of alternative patented drugs those deemed most cost-effective. Physicians were motivated to comply with formulary guidelines by persuasion, paperwork 35 Since reimbursement is to the pharmacist, the pharmacist dispensing a high-priced drug absorbs the cost difference.

1326

FM. Scherer

burdens justifying non-compliance, and in some cases financial penalties. As the use of formularies gained acceptance, health care organization drug procurement personnel realized that they could use the threat of a drug's exclusion from their formulary as a lever to elicit discounts from pharmaceutical manufacturers. Manufacturers of substitutable patented, branded drugs were played off against each other, and to an increasing degree they conceded substantial off-list discounts. By the early 1990s, HMOs were receiving discounts averaging 20 to 25 percent off the average wholesale prices paid by retail pharmacies [Boston Consulting Group (1993, Figure 1-3)]. A government study revealed that in 1991, the "best price" offered by a manufacturer to some customer usually an HMO or similarly bargaining hospital - carried a discount of 50 percent or more off the wholesale list price for 32 percent of all single-source (i.e., patented) drugs [U.S. Congressional Budget Office (1996, p. 28)]. For drugs dispensed on an inpatient basis within an HMO or hospital, there were several ways the negotiated discount could be realized. The organization could take delivery of the drug directly from the manufacturer and pay only the discounted price. But more frequently, health care organizations preferred to obtain their drug supplies from local wholesalers, who assumed the burden of maintaining inventories and assured same-day delivery of orders. In those instances, the wholesaler could bill the HMO at its standard wholesale price, after which the manufacturer would issue a rebate check to the HMO for the amount of the negotiated discount. Alternatively, the manufacturer could notify its wholesalers of its discount arrangements with individual HMOs, whereupon the wholesaler would deliver drugs to an HMO at the discounted price plus a negotiated wholesaling fee, receiving from the manufacturer a "chargeback" payment to cover the difference between the discounted price and the price at which the wholesaler had acquired its inventory from the manufacturer. PBMs entered the picture for the much larger volume of drugs obtained by patients from retail pharmacies. HMOs contracted with PBMs to reimburse retailers for insured prescriptions written by their affiliated physicians. When the HMO negotiated a discount with a manufacturer, the PBM would reimburse the retailer for its drug acquisition cost plus a negotiated dispensing fee, obtaining from the manufacturer a rebate (passed on the HMO) to cover the discount negotiated between the HMO and the manufacturer. (The negotiation of dispensing fees also imposed downward pressure on retail pharmacists' traditional 40 percent gross margins. 3 6 ) In addition, PBMs contracted to manage the retail pharmacy payment paperwork for companies with large numbers of employees whose health insurance included prescription drugs. To save money for their clients,

36 According to one account, retailers' margins on the sale of ethical drugs had been driven down by 1998

from 40 to 20 percent. "Why drugstore chains are in good health", Business Week, May 4, 1998, p. 170. See also Scherer (1997, pp. 245-246). In the late 1990s, PBMs and HMOs began concluding contracts under which pharmacy chains agreed to bill the PBM a fixed total price per prescription, regardless of the cost of the drug dispensed - an arrangement under which the pharmacies in effect bore the risk of dispensing unusually many high-cost drugs. Smaller retail pharmacies were wary of accepting such contracts because of the risks, and therefore lost HMO sales. See "Wrong Rx for their needs", Boston Globe, March 22, 1998, p. C1.

Ch. 25:


1327

PBMs established their own formularies, negotiated discounts with manufacturers for formulary inclusion, and assigned staff members to call physicians, urging them to prescribe lower-priced formulary drugs for covered patients when appropriate substitutes were available. These hard-ball tactics introduced new and powerful elements of competition into the pricing of patented drugs. A key lever used in negotiating discounts - exclusion of non-discounting drugs from formularies - may also have had more subtle adverse side effects. Before the tactic was widely adopted by HMOs, U.S. state government agencies frequently excluded new and high-priced drugs from formularies listing drugs available for Medicaid reimbursement. Statistical analyses reveal that expenditures on drugs were in fact reduced by such policies, but that hospital admission and similar costs may have increased by a more than offsetting amount as a consequence of using less effective drugs. See, e.g., Dranove (1989), Moore and Newman (1993), and, concerning flat limits on the number of reimbursable prescriptions per patient, Soumerai et al. (1991). The transfer of substantial control over prescription drug choices from individual physicians to formulary committees also elicited changes in pharmaceutical manufacturers' marketing strategies. Dispatching hordes of "detail persons" to make regular sales calls on individual physicians became less cost-effective, and so detailing budgets were cut sharply. To instigate patient pressure on physicians and from them on formulary committees, manufacturers in the United States began devoting substantial sums to newspaper and television advertising aimed directly at end consumers. Another unanticipated consequence of the manufacturer - HMO bargaining was an antitrust law suit that eventually pitted more than 25,000 retail pharmacists against 25 leading U.S. branded drug manufacturers. 3 7 The pharmacists alleged that by granting discounts to the HMOs who could threaten them with market share losses but not to the retail pharmacists, the drug manufacturers engaged in illegal price discrimination. They alleged also that discounts were withheld from the retailers through an illegal conspiracy, inferred inter alia from the cooperation between manufacturers and wholesalers to create a uniform computerized system for keeping wholesalers informed of negotiated discounts and paying chargebacks to the wholesalers who delivered discounted drugs to HMOs. Most of the plaintiffs and 13 drug companies agreed in 1996 to a settlement under which the manufacturers would pay damages of $351 million and offer discounts to retail pharmacy buying groups that could demonstrate "an ability to affect market share" through their own formulary and physician contact activities. 38 Four other companies settled for a total of $345 million in July 1998. In December 1998, the presiding judge

37 Inre Brand Name Prescription Drugs Antitrust Litigation, MDL-997, Master file no. 94 C 897 (consolidated before the U.S. Federal Court for the Northern District of Illinois). A symposium in the International

Journal of the Economics of Business, Vol. 4 (November 1997), joins the issues inarticles by eight economists who had consulted with either plaintiffs or respondents in the litigation. 38 Memorandum opinion of Judge Charles P. Kocoras, MDL-997, June 21, 1996. See also the Seventh Circuit Court of Appeals decision of August 15, 1997, which rejected many of Judge Kocoras' findings but did not overturn the settlement.

1328

FM. Scherer

dismissed the remaining conspiracy claims as "far fetched, improbable, and unreasonable". A U.S. government initiative seeking to reduce Medicaid drug costs had still another unanticipated consequence, restricting the size of the discounts manufacturers were willing to cede to hard-bargaining HMOs and other health care organizations. Under laws enacted in 1990 and extended thereafter, manufacturers were required to give the government agencies reimbursing Medicaid prescription drug purchases discounts equal to the best discount offered any non-governmental purchaser on any given product, but not less than 15.1 percent. As a consequence, manufacturers confronted with demands for deep discounts from a strong HMO realized that if they yielded to the HMO, they would also have to increase their discounts on Medicaid sales, which in 1995 amounted to 17.6 percent of all out-patient prescription drug sales [PhRMA (1997, p. 35)]. Two statistical studies have shown that the perverse most-favored customer incentives generated by the Medicaid discount law substantially reduced the discounts received by other health care organizations [U.S. Congressional Budget Office (1996, pp. 27-44), and Morton (1997a)]. 8. Profits and price controls Pharmaceuticals are sold under classic monopolistic competition conditions. Patents and product differentiation lead to prices that are well above production costs. But companies strive for partial monopoly positions and high margins by introducing new drugs. To do so, they incur substantial R&D costs and marketing costs, reducing bottom-line profits. When superior new products emerge from drug R&D, consumers and producers alike benefit. Whether high expenditures on sales promotion are in the net socially beneficial is more heatedly debated. On one hand, information is disseminated to physicians and (more recently) consumers, speeding the entry of new drugs into medical practice. On the other hand, as a governmental commission concluded, "It is doubtful ... that ... detail men ... give invariably unprejudiced and objective advice". [Task Force on

Prescription Drugs (1969, pp. 9-10)]. Compare also Schwartzman (1976, Chapter 9), Comanor (1986, pp. 1196-1199), Beales (1996), and Kopp and Sheffet (1997). In the United States, the pharmaceutical industry has for decades appeared at or near the top of industry rankings by after-tax profit returns on stockholders' equity [Scherer (1996, p. 342)]. However, profit reports prepared following conventional accounting practices, including current-year writeoff of research and development expenditures, tend to overstate true economic profitability, given the growth rates experienced by pharmaceutical firms. When drug makers' R&D outlays were capitalized and amortized at plausible rates, the industry's overall rate of return on invested capital in the 1970s and 1980s was found to exceed all-industry averages by only two to three percentage points [U.S. Office of Technology Assessment (1993, Chapter 4)]. The perception, correct or incorrect, that pharmaceutical prices and profits have been excessive, the taxpayer burdens from rising public health care costs, and the belief, especially in smaller nations, that reducing drug prices and profits will at best have

Ch. 25:


1329

a minor impact on R&D expenditures by companies oriented toward serving worldwide markets, have led many governments to impose more or less thoroughgoing price controls on pharmaceutical products. Among 56 nations whose governmental policies toward the pharmaceutical industry were surveyed by Ballance et al. (1992, pp. 140145 and 166-171), 30 nations (12 industrialized and 18 developing) were characterized as having "substantial" price controls and 20 (11 industrialized and nine developing) as having "limited" controls. Only six (all in the developing category) were said to have no controls. There are countless variations in the ways governments regulate drug prices. These can be compacted into five broad groupings - reference pricing, item-by-item negotiation and control, formula pricing, profit or rate of return regulation, and capping or budgetary constraint controls. See Danzon (1997), Shulman and Lasagna (1994), and U.S. Office of Technology Assessment (1993, pp. 250-262), on which the next seven paragraphs are based. Many nations' policies entail a mix of the various methods, with the mix changing over time, so what follows can provide only selective snapshots. Under reference pricing, more-or-less comparable drugs are placed into a reference group, and reimbursement is provided under national or regional health insurance plans only at the lowest price within the reference group. The U.S. maximum allowable cost approach to Medicaid drug reimbursement is a relatively innocuous version, placing generics and branded drugs with identical active ingredients in the same group. A more drastic approach was taken by Germany beginning in 1989 and Sweden beginning in 1993. Different chemical entities treating the same illness are placed in the same reference group. The broader the reference group and the more it includes new formulations along with older drugs, the more likely it is to discourage investments in discovering and developing superior new drugs. In France and (until 1993) Italy, the prices of individual outpatient drugs seeking reimbursement under national' health insurance plans were set in administrative proceedings taking into account a wide array of criteria, including therapeutic novelty and contribution to the economy. Drugs produced and developed locally tended to receive higher prices than imported drugs, which created incentives for local firms to develop and introduce numerous new drugs of insufficient therapeutic novelty to achieve significant sales outside the home market. This, along with the low standards imposed by the agencies regulating new drug introductions, helps explain the relatively modest external sales of French and Italian drug manufacturers, as shown in Figure 3. Under the health care reforms proposed in 1993 by U.S. President Clinton but rejected by the Congress, a different form of ad hoc regulation was contemplated. An Advisory Council on Breakthrough Drugs was to be charged with reviewing the prices of new drugs and, in cases where they were considered excessive, implementing measures ranging from public suasion or "jawboning" to making the drugs ineligible for health insurance reimbursement. The drugs most likely to be singled out for this regulation were the "blockbusters" in the first decile of Figure 4. The difficulty with this approach, obvious from Figure 4, is that curbing significantly the prices and profits of blockbuster drugs could make it difficult for companies to recover their research and development

1330

EM. Scherer

investments on less successful drugs - i.e., in the fourth through tenth deciles. Severe impairment of R&D incentives could result. Many nations, including Italy since 1993 [Fattore (1996)] and Canada since 1987, relate the reimbursable prices of relatively new drugs to the prices of the same drugs in other nations. When nations (such as Spain) characterized by generally low prices are included in the comparison group, this creates incentives for multinational drug manufacturers to set prices higher in the comparison group jurisdiction than those they would otherwise be inclined to charge. 3 9 Or in the case of India, where both low incomes and regulation limit sustainable prices, multinationals are said sometimes not to market drugs at all until late in their life cycle so that low prices in India cannot be cited to reduce prices in other nations making international comparisons. See Lanjouw (1997). In Japan, the vast majority of drugs are dispensed directly by physicians, who are then reimbursed by government health authorities on a formula basis for the drugs. New drugs receive relatively high prices, and after that, their prices are reduced downward systematically with the drug's age. This system has two important incentive effects. For one, to encourage the use of their drugs, manufacturers set prices that allow physicians generous margins between the physician's acquisition cost and the reimbursed price, leading to the extraordinarily high prescribing rates observed in Japan. Second, because new drugs command the highest prices, manufacturers have strong incentives, as in France and Italy, to introduce many new drugs, whether or not they make significant therapeutic contributions. This in turn is partly responsible for the poor external market performance of Japanese drug manufacturers evident in Figure 3. See Thomas (1996) and Ikegami et al. (1998). The United Kingdom is the only nation known to have a rate of return regulation system analogous to the way electrical and telephone utilities were regulated in the United States for many decades. In an annual determination, the assets of individual companies, including the capitalized value of research and development outlays, are measured. Each company negotiates with the regulatory authority an allowed beforetax rate of return on its assets, usually in the range of 17 to 21 percent. Prescription drug sales revenues are set (or adjusted after-the-fact) so that, after operating, R&D, and sales promotion costs are deducted, the company is left with profit sufficient to yield the agreed-upon rate of return on assets. In the cost calculations, promotional expenditures can be deducted only up to a limit of approximately 9 percent of sales. The U.K. "Price Regulation Scheme" would appear to reward investments in research and development

39 Wide differences in drug price levels among the member nations of the European Union led to further complications. Wholesalers in low-price nations such as Spain attempted to ship drugs imported from, e.g., Germany back to Germany to arbitrage the higher prices there. When manufacturers sought to curb such "parallel imports", the Common Market authorities intervened, charging illegal restraint of competition. See the Commission decision In re Adalat, Case IV/34.279/F3, 1995. For diverse views on policy toward parallel imports and arguments favoring inter-national price discrimination as a "second best" Ramsey pricing solution to the problem of recovering fixed R&D costs, see Yarrow (1995), Danzon (1997, Chapter 7), Bangemann (1997), and Anis and Wen (1998).

Ch. 25:

The PhannaceuticalIndustry

1331

and hence to avoid the negative incentive problems in many other nations' regulatory approaches. However, there is a paradox. If the scheme is implemented mechanically, large companies with R&D portfolios containing many projects tend to realize substantially higher returns on investment than small companies with few projects. See Scherer (1995, pp. 36-38). Given the high skewness of drug development project outcomes revealed by Figure 4, companies with many projects can include the substantial R&D investment from numerous "losers" as well from (the few) blockbusters in their R&D asset base, and the large investment base will allow the companies to realize most, if not all, of the profit potential from blockbusters. If a small company is lucky enough to develop a blockbuster, it will by its very smallness have few losers in its investment base, so the revenues it is allowed to realize on the blockbuster will be severely limited by regulation. If on the other hand (with appreciable probability) it achieves no blockbuster, its returns will be severely limited by market competition. Germany illustrates the use of aggregate budget constraints and roll-backs. In an attempt to control escalating health care costs, the Federal Health Ministry beginning in 1993 set a tight overall drug budget, requiring inter alia a roll-back from previous spending levels. The first DM 280 million of spending above that target was to be deducted from the incomes of physicians. If the budget was exceeded by DM 281-360 million, the excess was to be deducted from reimbursements to drug manufacturers.4 0 Between 1995 and 1997, German drug budgets were decentralized regionally out to the level of individual physicians (as is also done in the United Kingdom). An apparent consequence of individual physician spending constraints was that primary care physicians referred increased numbers of patients to specialists and hospitals, who were subject to different individual constraints [Sch6ffski (1996)]. In 1998, cost containment emphasis in Germany shifted away from drug budget constraints toward increased individual patient co-payments. In sum, efforts by national authorities to curb pharmaceutical costs and offset the demand-increasing effects of generous health care insurance by imposing drug price controls are found throughout the industrialized and less-developed world. These sometimes succeed in their proximate goal, but cause bulges in other parts of the health care balloon, bias new drug research and development incentives, and distort international trade and investment patterns. Although one may share the underlying cost control goals, a review of the consequences suggests that the aversion of most economists to price controls is well-founded.

9. Conclusion The pharmaceutical industry has made enormous contributions to health care in the half century since World War II as the drug research and development revolution gained

40 France introduced manufacturer-specific budget targets in 1994.

1332

EM. Scherer

momentum. Progress in biological science and molecular engineering is likely to provide the basis for further dramatic therapeutic advances in the future. But the conditions that create strong incentives for investment in pharmaceutical R&D also arouse public concern over monopoly positions, high prices, and the introduction of products of uncertain efficacy or safety. From that concern flow regulatory interventions into clinical testing protocols and pricing that could retard future technological progress. The problem is complicated by the fact that individual nation states can rationally behave as free riders or, more accurately, cheap riders, ignoring the consequences of their policies on drug R&D decisions in other parts of a complex multinational industry. Achieving the right tradeoff between progress, affordability, and optimal provision of test information remains an elusive goal.

References Ager, B. (1997), "Patents, pharmaceuticals and generics", Eurohealth 2:35-36. Anis, A.H., and Q. Wen (1998), "Price regulation of pharmaceuticals in Canada", Journal of Health Economics 17:21-38. Bakke, O.M., M. Manochia, F. de Abajo, K.I. Kaitin and L. Lasagna (1995), "Drug safety discontinuations in the United Kingdom, the United States, and Spain from 1974 through 1993", Clinical Pharmacology and Therapeutics, 58:108-117. Ballance, R., J. Pogany and H. Forstner (1992), The World's Pharmaceutical Industries: An International Perspective on Innovation, Competition and Policy (Edward Elgar for United Nations Industrial Development Organization, Hants, UK). Bangemann, M. (1997), "Completing the single pharmaceutical market", Eurohealth 3(1):22-23. Beales, J.J. (1996), "New uses for old drugs", in: R.B. Helms, ed., Competitive Strategies in the Pharmaceutical Industry (AEI Press, Washington) 281-305. Berndt, E.R., L. Bui, D.R. Reiley and G.L. Urban (1995), "Information, marketing, and pricing in the U.S. antiulcer drug market", American Economic Review 85(2):100-105. Bond, R.S., and D.F. Lean (1977), Sales, Promotion, and Product Differentiation in Two Prescription Drug Markets (Federal Trade Commission, Washington). Boston Consulting Group (1993), The Changing Environment for U.S. Pharmaceuticals (Boston). Caves, R.E., M. Whinston and M. Hurwitz (1991), "Patent expiration, entry, and competition in the U.S. pharmaceutical industry", Brookings Papers on Economic Activity: Microeconomics 1991:1-48. Cockbum, I., and R. Henderson (1998), "Absorptive capacity, coauthoring behavior, and the organization of research in drug discovery", Journal of Industrial Economics 46:157-182. Cohen, W.M., R.R. Nelson and J. Walsh (1997), "Appropriability conditions and why firms patent and why they do not in the American manufacturing sector", Working Paper (Carnegie-Mellon University, Pittsburgh). Comanor, W.S. (1986), "The political economy of the pharmaceutical industry", Journal of Economic Literature 24:1178-1217. Cooper, J.D. (1970), "The sources of innovation", in: J.D. Cooper, ed., The Economics of Drug Innovation (American University, Washington) 41-62. Danzon, P.M. (1997), Pharmaceutical Price Regulation (AEI Press, Washington). Deardorff, A.V. (1992), "Welfare effects of global patent protection", Economica 59:35-51. DiMasi, J.A., R.W. Hansen, H.G. Grabowski and L. Lasagna (1991), "Cost of innovation in the pharmaceutical industry", Journal of Health Economics 10:107-142. DiMasi, J.A., M. Seibring and L. Lasagna (1994), "New drug development in the United States from 1963 to 1992", Clinical Pharmacology and Therapeutics 55:609-622.

Ch. 25:


1333

Dorfman, R., and P.O. Steiner (1954), "Optimal advertising and optimal quality", American Economic Review 44:826-836. Dranove, D. (1989), "Medicaid drug formulary restrictions", Journal of Law and Economics 32:143-162. Ellison, S.F., I. Cockburn, Z. Griliches and J. Hausman (1997), "Characteristics of demand for pharmaceutical products: An examination of four cephalosporins", RAND Journal of Economics 28:427-446. Fattore, G. (1996), "Pharmaceutical policy in Italy 1992-1995", Eurohealth 2:34-35. Frank, R.G., and D.S. Salkever (1992), "Pricing, patent loss and the market for pharmaceuticals", Southern Economic Journal 59:165-179. Frank, R.G., and D.S. Salkever (1997), "Generic entry and the pricing of pharmaceuticals", Journal of Economics and Management Strategy 6:75-90. Gambardella, A. (1995), Science and Innovation: The US Pharmaceutical Industry During the 1980s (Cambridge University Press, Cambridge, UK). Garber, S. (1993), "Product liability and the economics of pharmaceuticals and medical devices", RAND Corporation Report R-4285-ICJ (Santa Monica, CA). Glied, S. (2000), "Managed care", in: A.J. Culyer and J.P. Newhouse, eds., Handbook of Health Economics (Elsevier, Amsterdam) Chapter 13. Gorecki, P.K. (1986), "The importance of being first: The case of prescription drugs in Canada", International Journal of Industrial Organization 4:371-396. Grabowski, H.G. (1976), Drug Regulation and Innovation (American Enterprise Institute, Washington). Grabowski, H.G., and J.M. Vernon (1983), The Regulation of Pharmaceuticals (American Enterprise Institute, Washington). Grabowski, H.G., and J.M. Vernon (1990), "A new look at the returns and risks to pharmaceutical R&D", Management Science 36:804-821. Grabowski, H.G., and J.M. Vernon (1992) "Brand loyalty, entry, and price competition in pharmaceuticals after the 1984 drug act", Journal of Law and Economics 35:331-350. Grabowski, H.G., and J.M. Vernon (1994), "Returns on new drug introductions in the 1980s", Journal of Health Economics 13:383-406. Grabowski, H.G., and J.M. Vernon (1996), "Longer patents for increased generic competition in the US", PharmacoEconomics 10(Supplement 2): 110-123. Grabowski, H.G., J.M. Vernon and L.G. Thomas (1978), "Estimating the effects of regulation on innovation: An international comparative analysis", Journal of Law and Economics 21:133-163. Griliches, Z., and I. Cockburn (1993), "Generics and new goods in pharmaceutical price indexes", Working Paper No. 4272 (National Bureau of Economic Research). Henderson, R., and I. Cockburn (1996), "Scale, scope, and spillovers: The determinants of research productivity in drug discovery", RAND Journal of Economics 27:32-59. Heppell, S. (1996), "The new European system for regulating medicinal products", Eurohealth 2:28-29. Holmes, J., and J. Dunning (1995), "Factors influencing the location of multinational investment in the pharmaceutical industry", in: A. Towse, ed., Industrial Policy and the Pharmaceutical Industry (Office of Health Economics, London) 92-105. UT Research Institute (1968), Technology in Retrospect and Critical Events in Science (Illinois Institute of Technology, Chicago). Ikegami, N., S. Ikeda and H. Kawai (1998), "Why medical care costs in Japan have increased despite declining prices of pharmaceuticals", PharmacoEconomics 14(Supplement 1):97-105. Kaitin, K.I. (1997), "The new drug approvals of 1993, 1994, and 1995", American Journal of Therapeutics 4:46-54. Kitch, E.W. (1973), "The patent system and the new drug application", in: R.L. Landau, ed., Regulating New Drugs (University of Chicago Center for Policy Study) 81-108. Kopp, S.W., and M.J. Sheffer (1997), "The effect of direct-to-consumer advertising of prescription drugs on retail gross margins", Journal of Public Policy and Marketing 16:270-276. Lanjouw, J.O. (1997), "The introduction of pharmaceutical product patents in India", Discussion Paper No. 775 (Yale University Economic Growth Center).

1334

FM. Scherer

Leibowitz, A., W.G. Manning and J.P. Newhouse (1985), "The demand for prescription drugs as a function of cost sharing", Social Science Medicine 21:251-277. Levin, R.C., A.K. Klevorick, R.R. Nelson and S.G. Winter (1987), "Appropriating the returns from industrial research and development", Brookings Papers on Economic Activity 3:783-821. Lichtenberg, FR. (1996), "The effect of pharmaceutical utilization and innovation on hospitalization and mortality", Working Paper No. 5418 (National Bureau of Economic Research, Cambridge, MA). Lichtenberg, F.R. (1998), "Pharmaceutical innovation, mortality reduction, and economic growth", Working Paper No. 6569 (National Bureau of Economic Research, Cambridge, MA). Lu, .L., and W.S. Comanor (1998), "Strategic pricing of new pharmaceuticals", Review of Economics and Statistics 80:108-118. Mann, C.C., and M.L. Plummer (1991), The Aspirin Wars (Harvard Business School Press, Boston). Manning, R.L. (1994), "Changing rules in tort law and the market for childhood vaccines", Journal of Law and Economics 37:247-275. Manning, R.L. (1997), "Products liability and prescription drug prices in Canada and the United States", Journal of Law and Economics 49:203-244. Mansfield, E. (1970), "Comment", in: J.D. Cooper, ed., The Economics of Drug Innovation (American University, Washington) 149-154. Mansfield, E. (1995), "Academic research underlying industrial innovations", Review of Economics and Statistics 77:55-65. Masson, A., and R.L. Steiner (1985), "Generic substitution and prescription drug prices", Staff Report (Federal Trade Commission, Washington). McFetridge, D.G. (1997), "Intellectual property rights and the location of innovative activity: The Canadian experience", Working Paper (Carleton University, Ottawa). McRae, J.J., and F. Tapon (1985), "Some empirical evidence on post-patent barriers to entry in the Canadian pharmaceutical industry", Journal of Health Economics 4:43-61. Moore, W.J., and R. Newman (1993), "Drug formulary restrictions as a cost-containment policy in Medicaid programs", Journal of Law and Economics 36:71-114. Morton, F.S. (1997a), "The strategic response by pharmaceutical firms to the Medicaid most-favoredcustomer rules", RAND Journal of Economics 28:269-290. Morton, F.S. (1997b), "The objectives of the FDA's Office of Generic Drugs", Working Paper No. 6143 (National Bureau of Economic Research). Neary, I. (1995), "Japanese industrial policy and the pharmaceutical industry", in: A. Towse, ed., Industrial Policy and the Pharmaceutical Industry (Office of Health Economics, London) 12-25. Newhouse, J.P., and the Insurance Experiment Group (1993), Free for All? Lessons from the RAND Health Insurance Experiment (Harvard University Press, Cambridge). Peltzman, S. (1973), "An evaluation of consumer protection legislation: The 1962 drug amendments", Journal of Political Economy 81:1049-1091. Peltzman, S. (1987), "The health effects of mandatory prescriptions", Journal of Law and Economics 30:207238. Perry, G. (1996), "Generic medicines: How to ensure their effective contribution to health care", Eurohealth 2:32-34. PhRMA (1997), Industry Profile: 1997 (Pharmaceutical Research and Manufacturers of America, Washington). PhRMA (1998), Industry Profile: 1998 (Pharmaceutical Research and Manufacturers of America, Washington). Rizzo, J.A. (1997), "Advertising and competition in the ethical pharmaceutical industry: The case of antihypertensive drugs", Working Paper (Yale University). Robinson, W.T., G. Kalyanaram and G. Urban (1994), "First-mover advantages from pioneering new markets: A survey of the empirical evidence", Review of Industrial Organization 9:1-24. Ryan, M.P. (1998), Knowledge Diplomacy: Global Competition and the Politics of Intellectual Property (Brookings, Washington).

Ch. 25:


1335

Santoro, M. (1992), "Pfizer: Protecting intellectual property in a global marketplace", Case Study N9-392-072 (Harvard Business School, Boston). Schachtel, B.P. (1994), "Sensible and sensitive clinical trials in the evaluation of over-the-counter drugs", in: S.R. Shulman, ed., Moving Prescription Drugs to the Over-the-Counter Market (Tufts University Human Nutrition Center, Boston). Scherer, EM. (1995), "US industrial Policy in the pharmaceutical industry", in: A. Towse, ed., Industrial Policy and the Pharmaceutical Industry (Office of Health Economics, London) 26-39. Scherer, F.M. (1996), Industry Structure, Strategy, and Public Policy (Harper-Collins, New York). Scherer, EM. (1997), "How U.S. antitrust can go astray", International Journal of the Economics of Business 4:239-256. Scherer, EM. (1998), "The size distribution of profits from innovation", Annales d'Economie et de Statistique 49/50:496-516. Scherer, EM., and D.R. Ross (1990), Industrial Market Structure and Economic Performance, 3rd edn. (Houghton-Mifflin, Boston). Scherer, EM., and S. Weisburst (1995), "Economic effects of strengthening pharmaceutical patent protection in Italy", International Review of Industrial Property and Copyright Law 26:1009-1024. Sch6ffski, 0. (1996), "Consequences of implementing a drug budget for office-based physicians in Germany", PharmacoEconomics 10(Supplement 2):37-47. Schwartzman, D. (1976), Innovation in the Pharmaceutical Industry (Johns Hopkins University Press, Baltimore). Schwartzman, D. (1997), "Has pharmaceutical research become more scientific?", Review of Industrial Organization 11:841-851. Shulman, S.R., and K.I. Kaitin (1996), "The prescription drug user fee act of 1992", PharmacoEconomics 9(2):121-133. Shulman, S.R., and L. Lasagna, eds. (1994), Pricing and Reimbursement of Pharmaceuticals: An Evaluation of Cost-Containment Strategies, Symposium proceedings, PharmacoEconomics 6(Supplement 1). Soumerai, S.B., D. Ross-Degnan, J. Ahom, T.J. McLaughlin and I. Choodnovskiy (1991), "Effects of Medicaid drug-payment limits on admission to hospitals and nursing homes", New England Journal of Medicine 325:1072-1077. Steiner, R.L. (1993), 'The inverse association between the margins of manufacturers and retailers", Review of Industrial Organization 8:717-740. Task Force on Prescription Drugs (1968), The Drug Makers and the Drug Distributors (US Department of Health, Education and Welfare, Washington). Task Force on Prescription Drugs (1969), Final Report (US Department of Health, Education and Welfare, Washington). Temin, P. (1979a), "The origin of compulsory drug prescriptions", Journal of Law and Economics 22:91-105. Temin, P. (1979b), "Technology, regulation, and market structure in the modem pharmaceutical industry", Bell Journal of Economics 10:429-446. Thomas, L.G. (1996), "Industrial policy and international competitiveness in the pharmaceutical industry", in: R.B. Helms, ed., Competitive Strategies in the Pharmaceutical Industry (AEI Press, Washington) 107-129. Toole, A.A. (1997), "The impact of federally funded basic research on industrial innovation: Evidence from the pharmaceutical industry", Working Paper (Christensen Associates, Madison, WI). Tufts University Center for the Study of Drug Development (1997), "Tufts CSDD testifies before Canadian Parliamentary Committee about U.S. experience with patent extension", Tufts CSDD Newsletter 22:1-2. US Bureau of the Census (1992), 1987 Census of Manufactures, Concentration Ratios in Manufacturing, MC87-S-6 (USGPO, Washington). US Bureau of the Census (1996), Statistical Abstract of the United States: 1996 (USGPO, Washington). US Bureau of the Census (1997), 1995 Annual Survey of Manufactures, Value of Product Shipments, M95(AS)-2 (USGPO, Washington). US Congressional Budget Office (1996), How the Medicaid Rebate on Prescription Drugs Affects Pricing in the Pharmaceutical Industry (CBO Papers, Washington).

1336

EM. Scherer

US Congressional Budget Office (1998), How Increased Competition from Generic Drugs Has Affected Prices and Returns in the Pharmaceutical Industry (Washington). US Federal Trade Commission (1958), Economic Report on Antibiotics Manufacture (USGPO, Washington). US Federal Trade Commission (1985), Statistical Report: Annual Line of Business Report: 1977 (Washington). US House of Representatives, Committee on Science and Technology (1980), Report, The Food and Drug Administration's Process for Approving New Drugs (USGPO, Washington). US National Science Board (1998), Science and Engineering Indicators: 1998 (USGPO, Washington). US National Science Foundation (1996), Research and Development in Industry: 1993, NSF 96-304 (Washington). US Office of Technology Assessment (1984), Federal Policies and the Medical Devices Industry (USGPO, Washington). US Office of Technology Assessment (1993), Pharmaceutical R&D: Costs, Risks and Rewards (USGPO, Washington). Wardell, W., and L. Lasagna (1975), Regulation and Drug Development (American Enterprise Institute, Washington). Weintraub, M. (1994), "OTC drugs: Definitional and methodologic aspects", in: S.R. Shulman, ed., Moving Prescription Drugs to the Over-the-Counter Market (Tufts University Human Nutrition Research Center, Boston). Yarrow, G. (1995), "CED and EC member state industrial policy and the pharmaceutical industry", in: A. Towse, ed., Industrial Policy and the Pharmaceutical Industry (Office of Health Economics, London). Zuger, A. (1997), "Drug companies' sales pitch: Ask your doctor", New York Times, August 5, p. C1.