new evidence is based primarily on a comparative analysis of developments in the United States and United Kingdom. In particular, we attempt to separate the impact of increased regulatory controls in the United States (stemming from the 1962 amendments to the 1938 Federal Food, Drug, and Cosmetic Act)' from other factors by using the U.K. industry as a control. Since firms in the latter country have been governed by a very different regulatory system but are similar to U.S. firms in most other ways, we feel that comparative analysis is a very fruitful way of approaching this question. The paper has the following plan. First, as background to our analysis, we briefly describe the structural changes that have characterized new product innovation in ethical drugs, as well as the hypothesized relations which account for these trends. We then review two past empirical studies that have attempted to explain the most important and controversial of such structural changes: declining levels of new product introductions in the United States. Finally, a model previously developed by Martin Baily2 is reformulated and employed in a comparative analysis of the U.S. and U.K. industries.

I. STRUCTURAL CHANGES IN PHARMACEUTICAL INNOVATION:
TRENDS AND HYPOTHESES

Evidence from a number of studies indicates that the American pharmaceutical industry has undergone some fundamental shifts in innovational structure and performance over recent years. This section briefly documents these basic trends and more systematically considers the proliferating hypotheses which have been advanced to explain these structural changes. A. Trends in Pharmaceutical Innovation

In the post-1962 period, the U.S. pharmaceutical industry has experienced the following.

i) Declining Rates of New Product Introductions. This decline is illustrated in Figure I. It shows the total new chemical entities (NCEs) introduced annually into the United States over the period 1954-1974, as well as the subset of each year's introductions that were discovered in the United States by the pharmaceutical industry.3 NCEs are the most important cate

1 Federal Food, Drug, and Cosmetic Act of 1938, 52 Stat. 1040, c. 675 as amended by Pub. L. No. 80-625, 21 U.S.C. §§ 1-517 (1964).

2 Martin N. Baily, Research and Development Costs and Returns: The U.S. Pharmaceutical Industry, 80 J. Pol. Econ. 70 (1972).

3 Data on NCEs and their years of introduction were obtained from Paul de Haen, Inc. See note 54 infra. Biologicals and diagnostics were deleted from the analysis. Information on the country of discovery was also obtained from de Haen, as well as supplementary sources. An NCE is regarded as discovered in a particular country if the research laboratory producing the

Introductions and Discoveries of New Chemical Entities by Domestic Firms and Constant (1958) Dollar Expenditures on Pharmaceutical Research and Development, the United States (1954-1974).

gory of new products because they represent compounds not previously marketed and include all significant new therapeutic advances. Thus NCES form a reasonable index of innovative output. Other new products involve combinations of existing products, new dosage forms, or new brand names. In Table 1 data on NCE introductions are grouped into five-year periods beginning in 1957. The table shows that the rate of introductions over the most recent five-year period is less than one-third the rate prevailing in a similar period a decade ago. The third column of Table 1, which shows the total market shares captured by new NCEs over these three periods, underscores the extent to which new product innovation has declined as a competitive factor in the ethical drug market.

ii) Increasing Costs of Innovation. Over the same time frame in which introductions and discoveries of NCEs have significantly declined, industry R & D expenditures have increased severalfold. These trends imply a rather

entity was located in that country, irrespective of the nationality of laboratory ownership. See the Appendix for details on the procedures used in the text in this regard.

The choice of period here was dictated by the availability of sales data (no data were available prior to 1957) and the three-year average sales measure employed in Table 1. The sales data were obtained from Intercontinental Medical Statistics, Inc. See note 63 infra. The nature of these data is discussed in the Appendix.

TABLE 1

NUMBER AND Sales of NEW CHEMICAL ENTITIES

IN THE PRE- AND POSTAMENDMENT PERIOD IN THE UNITED STATES

Average annual sales of all NCEs introduced during this period as a percentage of total ethical drug sales in the last year of the period.

Sources: Lists of new chemical entities in each year were obtained from Paul de Haen, Annual New Product Parade, various issues, all information on ethical drug sales were obtained from Intercontinental Medical Statistics, various years.

formidable increase in the costs of producing an NCE, an increase which has been documented in studies by Clymer, Mund, and Sarett. In particular, Sarett suggests that over the decade 1962 to 1972, development costs per NCE rose from 1.2 to 11.5 million dollars.

iii) Increasing Risks for Innovation. In addition, there appears to be a corresponding increase in the risks and uncertainty associated with innovational activity. One measure of risk in this industry is the attrition rates for compounds that undergo clinical testing but fail to become commercial products. Clymer estimates that in the 1950s, the attrition rate of drugs undergoing clinical tests was two out of three. The best estimate of the current situation appears to be that less than one of every ten new compounds entering clinical trials become new products."

In short, the decline in new product outputs in the drug industry has been accompanied by a number of adverse structural trends on the input side of the innovational process. Total development time and costs have increased severalfold. Furthermore, innovation has become subject to greater risks and uncertainty. These adverse structural trends in both innovational inputs and outputs appear related to more fundamental underlying changes in the

3 Harold A. Clymer, The Changing Costs and Risks of Pharmaceutical Innovation, in The Economics of Drug Innovation 109 (Joseph D. Cooper ed. 1970); Vernon A. Mund, The Return on Investment of the Innovative Pharmaceutical Firm, in the Economics of Drug Innovation 125 (Joseph D. Cooper ed. 1970); L. H. Sarett, FDA Regulations and Their Influence on Future R & D, 17 Int'l J. Research Management 18 (1974).

Harold A. Clymer, supra note 5, at 152.

"In particular, Louis Lasagna & William M. Wardell, The Rate of New Drug Discovery, in Drug Development and Marketing 155 (R. B. Helms ed. 1975) (Am. Enterprise Inst.), present data (from a questionnaire survey of 15 large firms accounting for 80% of U.S. research) that indicate only 7.1% of all new drug investigational plans (INDs) filed by these firms between 196J and 1967 had become approved NCEs by April 1974 (the date of their study).

innovational process. A review of the hypothesized causes of these adverse trends follows.

B. The Hypotheses

i) Increased FDA Regulation. Of the five hypotheses mentioned in the introduction, the role of increased regulation associated with the 1962 Kefauver-Harris amendments has received the most prominent attention in explaining declining pharmaceutical innovation. The antecedent 1938 Food, Drug, and Cosmetic Act required all new drugs to undergo a premarket approval process based on safety. Under this law, the FDA also had to reject a new drug compound within a period of sixty days or the new compound was automatically approved for marketing by the manufacturer.

The 1962 Kefauver-Harris amendments extended the regulatory controls of the FDA in several ways. First, it required firms to submit documented scientific evidence on a new drug's efficacy as well as its safety. This led to a substantial increase in the number of tests that had to be performed and submitted to the FDA. Second, the FDA was given discretionary power over the clinical research process. Thus, prior to any testing in humans, firms must now submit a new drug investigational plan (IND) that provides the results of animal tests and plans for human testing. Third, the new regulations provided for FDA approval of advertising claims. Finally, the provision of automatic approval of a new drug application (NDA) after sixty days unless the FDA took specific action was effectively repealed.

Over the post-1962 period, therefore, there has been a significant increase in both the scope and intensity of regulatory controls on ethical drugs. As a consequence, it has been postulated that the costs of discovering and developing a new drug, along with the risks and uncertainty of drug innovation, have increased; and that this, in turn, has been a major factor in the observed decline in innovational output.

ii) Fewer Marginal and Ineffective Drugs. The initial response of the FDA to hypothesis (i) was to argue that the observed decline in pharmaceutical innovation is in fact illusory:

The relevant question is not and never has been how many new drugs are marketed each year, but rather how many significant, useful and unique therapeutic entities are developed. . . . The rate of development and marketing of truly important, significant, and unique therapeutic entities in this country has remained relatively stable for the past 22 years.

Unfortunately, it is difficult to substantiate this FDA claim as there is no list of important new drugs upon which there is general agreement by medi

Speech by Alexander Schmidt, The FDA Today: Critics, Congress, and Consumerism (Oct. 29, 1974 before the Nat'l Press Club, Wash., D.C.).

9

cal experts. Most lists from academic sources, for example, show a significant downward trend in important therapeutic advances, as does at least one prior FDA ranking of important new drugs. Furthermore, measures of pharmaceutical innovation based on economic criteria strongly suggest that a significant decline in real terms has occurred. The data presented in Table 1, in particular, indicate that the total market shares captured by NCES have declined over time in comparable fashion to the total number of NCE introductions. 10

Sam Peltzman has analyzed a related drug quality issue as to whether the large decline in NCE introductions could be explained by fewer ineffective drugs entering the marketplace after the 1962 amendments were passed. His analysis of data from three groups of experts-hospitals, panels employed by state public-assistance agencies, and the American Medical Association's Council on Drugs-does not support this view. These data suggest only a small fraction of the pre-1962 and post-1962 NCE introductions could be classified as ineffective."

In sum, the hypothesis that the observed decline in new product introductions has largely been concentrated in marginal or ineffective drugs is not generally supported by empirical analyses. Moreover, these data analyses show no real tendency for more recently introduced drugs to have either significantly higher average market shares or efficacy rates than those introduced in earlier periods.

iii) Depletion of Research Opportunities. More recently, the FDA (along with some prominent members of the biomedical community) have emphasized a very different hypothesis that the decline in pharmaceutical innovation is real, but that it is due to a depletion of research opportunities rather than increased regulation. This hypothesis has been described by former FDA Commissioner Schmidt as follows:

'Henry G. Grabowski, Drug Regulation and Innovation: Empirical Evidence and Policy Options (Am. Enterprise Inst. 1976).

10 Market measures are premised on the notion that drugs which obtain the largest shares do so because they offer consumers the most overall utility per dollar. One can argue, however, that some drugs which have important therapeutic properties, but for relatively rare diseases, will tend to obtain low market shares. In addition, market shares are presumably influenced not only by the therapeutic advance of a new drug but also by the innovating firm's market power. promotional strategies, and so forth. However, for the broad aggregate comparison presented above, these qualifications are not as important as they might be in other situations. This is because there is no reason to believe that these factors have changed markedly over time. especially not in a direction so as to produce the lower market shares for new drugs shown above. For example, it seems unlikely that the lower market shares can be plausibly accounted for by a shift toward the production of a relatively greater number of drugs for rare diseases. "In particular, these data suggest the incidence of ineffective new drugs was less than 10% in the pre- and post-1962 period. Peltzman also analyzes the growth rate patterns of NCEs in the pre- and post-1962 periods and argues they also support the findings of expert evaluations in this regard. See Sam Peltzman, An Evaluation of Consumer Protection Legislation: The 1962 Drug Amendments, 81 J. Pol. Econ. 1049, 1086 (1973).