Patent Protection and Pharmaceutical R&D Spending in Canada

Transcription

1 Patent Protection and Pharmaceutical R&D Spending in Canada Patent Protection and Pharmaceutical R&D Spending in Canada 29 BOHUMÍR PAZDERKA School of Business Queen s University Kingston, Ontario Après presque deux décennies de mise sous license obligatoire des médicaments vendus sous ordonnances, le Canada a restoré une protection complète des brevets suite à deux actions législatives entreprises en 1987 et Cet article analyse l impact du renforcissement des droits de propriété intellectuels sur les dépenses en R&D dans l industrie pharmaceutique canadienne. Les tendances des dépenses en R&D dans les industries canadiennes, les tendances des dépenses en R&D des industries pharmaceutiques de différents pays de même que les tendances dans la proportion des dépenses de R&D des multinationales américaines effectuées au Canada suggèrent une augmentation statistiquement significative dans les dépenses de R&D du secteur pharmaceutique au Canada après After almost two decades of compulsory licensing of prescription drugs, Canada restored full patent protection in two legislative steps taken in 1987 and The paper investigates the impact of the strengthening of intellectual property rights on corporate R&D spending in the Canadian pharmaceutical industry. Interindustry comparisons of R&D spending trends within Canada, intercountry comparisons of R&D spending trends within the pharmaceutical industry, as well as trends in Canada s share of foreign R&D spending of US-owned multinationals suggest a statistically significant increase in Canadian pharmaceutical R&D spending after INTRODUCTION Two legislative measures taken in 1987 and in 1992 by the Parliament of Canada dramatically strengthened the patent protection afforded to manufacturers of prescription drugs. This legislation repealed the 1969 amendments to the Patent Act which allowed for compulsory licensing of imports of prescription drugs. 1 Compulsory licensing facilitated the proliferation of cheaper generic substitutes for brand-name prescription drugs, and thus made possible substantial savings for consumers and for governments responsible for funding of drug costs. It also contributed in a major way to the development of a strong, mostly Canadian-owned, generic drug industry which by 1995 accounted for 10 percent of the dollar value of wholesale market and 21 percent of unit volumes of prescription drug sales (DFAIT 1996, p. 12). The brand-name sector of the industry has continued to be dominated by multinational companies,

2 30 Bohumír Pazderka mostly headquartered in the United States and Europe. In 1994, these companies, represented by the Pharmaceutical Manufacturers Association of Canada (PMAC), accounted for 78 percent of prescription drug sales, 70 percent of unit volumes, and approximately 80 percent of industry assets (DFAIT 1996, p. 12). The patent-holding multinational brand-name companies were affected by compulsory licensing in different degrees, depending on the importance of the licensed product(s) in their respective product portfolios. They objected to the licensing provision since its enactment, and publicized their reluctance to invest and conduct R&D in Canada. In an attempt to deal with the mounting objections to the compulsory licensing provisions from the multinational pharmaceutical industry and some segments of the Canadian scientific community, the government appointed a commission of inquiry (the Eastman Commission) to investigate the impact of compulsory licensing and suggest modifications. In its 1985 report, the commission concluded that the industry in Canada had been growing fairly steadily during the period, and that any negative impacts of the changes to the Patent Act in 1969 appear to have been more than offset by other factors like especially strong growth in demand (Eastman 1985, p. 65). For example, the commission could not detect any major adverse trend in R&D expenditures attributable to compulsory licensing. In order to enhance the incentives to conduct R&D in Canada, while at the same time encouraging price competition, the commission recommended a four-year period of exclusivity from generic competition, a variable royalty rate determined by the worldwide R&D-to-sales ratio of the patent holder, and the extension of patent protection to the product as well as process. A variety of improvements in the process of regulation, marketing, and distribution of drugs was also suggested. The pressure to modify the legislation gained additional impetus when compulsory licensing became an obstacle to successful completion of the Free Trade Agreement (FTA) between Canada and the United States. The review of the legislation culminated in amendments to the Patent Act, contained in Bill C-22, which came into effect on 7 December To facilitate its enactment, the PMAC undertook a public commitment to raise the R&D-to-sales ratio of the member companies from the 1984 level of 4.9 percent to 8 percent by 1991, and 10 percent by 1996 (PMAC 1993, p. 14). The legislation provided for a period of protection from compulsory licensing for ten years from the date of the first Notice of Compliance in the case of licence to import, and seven years in the case of licence to manufacture. Drugs invented and developed in Canada were protected for 20 years. The legislation also established the Patented Medicine Prices Review Board (PMPRB) with powers to review the prices of patented medicines and, if a price is found excessive, to direct the patentee to lower the price or permit a compulsory licence. It was also mandated to monitor the commitments made by the PMAC in respect of R&D spending. Complete abolition of compulsory licensing was proposed in Bill C-91, tabled on 23 June 1992, which came into effect on 4 February A parliamentary review of the legislation was mandated for 1997; it recommended against any substantial modifications. This paper focuses on the impact of the 1987 legislation on pharmaceutical industry R&D spending in Canada. While the modifications of patent legislation affect other aspects of the industry performance, among them the prices of prescription drugs, these are not dealt with here. The rest of the paper is organized as follows: the second section suggests an approach to evaluating the trends in Canadian pharmaceutical R&D spending against three types of benchmarks: (i) the R&D spending in other industries in Canada; (ii) R&D spending in pharmaceutical industries in other countries; and (iii) the trends in international distribution of R&D spending

3 Patent Protection and Pharmaceutical R&D Spending in Canada 31 by US-based multinational companies. Section three considers the confounding influence of a change in tax support for R&D, and the conclusion follows. IMPACT OF PATENT PROTECTION ON PHARMACEUTICAL R&D SPENDING Methodology An obvious way to assess the consequences of increased patent protection is to observe the trends in pharmaceutical R&D spending in Canada over time, and determine whether there exists a distinct (upward) change in the trend in R&D spending after 1987, the year of enactment of Bill C-22 (not enough data has been accumulated to evaluate the impact of Bill C-91). The timing of the impact is unlikely to be precise, due to the leads and lags in the corporate R&D planning process, and the possibility that some companies anticipated the passage of the legislation and modified their spending earlier. Any observed increase in pharmaceutical R&D spending in Canada in the post-1987 period could also be due to some underlying change in the general pattern of Canadian R&D spending, unrelated to modification of the patent legislation. It may be the case, for example, that R&D spending after 1987 accelerated in all (or most) Canadian industries as a result of improvements in R&D tax credits, or any number of other aspects of the general business climate. To deal with this problem, this paper relates Canadian pharmaceutical R&D spending to the R&D spending in other Canadian industries, in an attempt to establish whether the pharmaceutical industry exhibits a trend after 1987 that differs from other Canadian industries. Yet another possibility is that Canadian pharmaceutical R&D spending increased as a result of a paradigm change in pharmaceutical research, which took place in the 1980s. One such change may have been the shift from a largely empirical industrial research process (based on trials of many compounds) to a more rational search for innovation, grounded on a more effective use of scientific knowledge and computerized research technologies (Gambardella 1995). The ANBERD database, created by the OECD, is employed to compare the behaviour of R&D spending in the pharmaceutical industry in Canada in relation to other OECD countries. An attempt is made here to determine whether the pharmaceutical industry in Canada merely follows a general trend or whether it has behaved differently from other countries after the enactment of Bill C-22. Finally, an observed increase in R&D spending by Canadian subsidiaries of multinational pharmaceutical corporations could be due to a trend toward decentralization of R&D activities from corporate headquarters to subsidiaries (Grandstrand et al. 1993). This hypothesis is examined with the help of data on the position of Canada over time as a location for foreign investment and R&D activity of USowned multinationals, comparing pharmaceuticals with other industries. Observed Trends in the Canadian Pharmaceutical R&D Spending The PMAC has collected and reported data on R&D spending of its member companies for several decades, but reliable time-series based on consistent definitions are available only since Statistics Canada publishes R&D spending data on the industry as a whole; these are discussed in the next section. Since 1988, the PMAC has relied on the data collected by the PMPRB, which is mandated to monitor the R&D spending of patent-holding companies (the two sets of companies do not correspond exactly, but the difference in total R&D spending is relatively small). As shown in Table 1, there has been a marked acceleration in annual rates of growth of R&D spending for the years 1988 and Starting in 1990, the growth rates have exhibited a more moderate pace. 2 The R&D-to-sales ratio for all patent-holding companies monitored by the PMPRB rose from 6.1 percent in 1988 to 11.5 percent in 1997; for the PMAC member companies alone (last column in Table 1), the ratio rose from 6.5 percent to 12.9 percent over the same period (PMPRB 1998,

4 32 Bohumír Pazderka TABLE 1 Pharmaceutical R&D Expenditures in Canada, PMAC, (Canadian Dollars) Year Current $ Rate of Growth Const $ Rate of Growth Annual Rate of Growth R&D-to-Sales (million) (1984=100) (million) (1984=100) (constant $) Ratio N.A N.A N.A N.A Notes: 1. The R&D data for are from PMAC (1995b, p. 11), and represent R&D spending by PMAC member companies. The R&D data for are from PMPRB (1998, p. 28), and represent R&D spending by companies with active Canadian patents pertaining to a medicine sold in Canada. The PMAC (1995b, p. 11) reports an identical set of figures for the post-1988 period and describes them as R&D spending by PMAC member companies. The difference between companies with active Canadian patents and PMAC member companies is, however, very small. 2. The R&D expenditures reported here are those that would have been eligible for Investment Tax Credit for scientific research and experimental development under the provision of the Income Tax Act in effect on 1 December They include current expenditures (about 96 percent of the total), capital equipment expenditures, and allowable depreciation expenses. 3. The GDP implicit price index, 1986=100 (Statistics Canada Cat. No and ), was used to calculate R&D spending in constant dollars. 4. The R&D-to-sales ratios are for patent-holding PMAC member companies only, as reported by the PMPRB (1998, p. 28). p. 28). 3 The PMAC commitment to achieve a 10- percent R&D-to-sales ratio by 1996 was already satisfied in Apart from the volume of R&D spending, its composition has also been the subject of some concern. For example, Heller (1995, p. 252) and others believe that only research performed in universities and hospitals, rather than research performed by multinational drug companies, is likely to result in Canadian patents and thus make a long-term contribution to the Canadian economy. The history of the Canadian policy debate, however, also shows concern for the relatively low proportion of Canada s R&D performed in the private sector (see e.g., Palda 1993, pp ).

5 Patent Protection and Pharmaceutical R&D Spending in Canada 33 Breakdowns provided by the PMPRB (Annual Reports, various years) suggest that the bulk of the Canadian pharmaceutical R&D is performed by the patentees or by related companies (their share was 60.4 percent in 1988, dropped to 53.1 percent in 1990, reached 62.7 percent in 1992, and dropped to 55.3 percent in 1997). The share of universities and hospitals over the period fluctuated between 21 and 25 percent. The rest of the R&D was performed by companies other than the patentholder, by organizations such as private clinics, and by federal and provincial governments. It is impossible to determine whether this mix is optimal for a country the size of Canada. Applied research 4 represents the largest share of total R&D spending (it reached the high of 67.2 percent in 1988 and 1989, dropped to 57.1 percent in 1992, and rose again to 62 percent in 1997). Basic research as a share of total R&D reached its peak in 1990 (27.2 percent) and declined to 20.7 percent in 1997 (PMPRB, Annual Reports, various years). Some critics of Bills C-22 and C-91 have argued that any increase in R&D spending in Canada by foreign-owned multinationals which such legislation may bring about will most likely occur at the applied end of the spectrum (primarily clinical testing), and have alleged that it would generate fewer long-term benefits than would result from basic R&D. These conjectures about the relative merits of basic versus applied research have not, however, been empirically documented. R&D in Pharmaceuticals versus other Canadian Industries The growth in pharmaceutical R&D spending in Canada in the period, reported in Table 1, may have been affected by tighter patent protection, but may have also resulted from other factors, such as an improved general climate for R&D in Canada. Since these factors affect the whole economy, other industries and sectors should exhibit a similar change in R&D spending growth. The R&D expenditures of selected Canadian industry aggregates, as reported by Statistics Canada, are presented in Table 2. The pharmaceutical R&D as a share of all manufacturing R&D spending rose from 4.5 percent in 1975 to 10.4 percent in The ratio of pharmaceutical R&D to the total of R&D spending in all industries rose from 3.6 percent to 6.3 percent over the same period. R&D spending in all industries does, of course, fluctuate from year to year. The ratios calculated in the last two columns of Table 2 are thus affected by fluctuations in both the numerator and in the denominator. 5 In order to deal with random deviations and to determine whether a structural change in trend occurred around the time Bill C-22 was enacted, a statistical test was performed with the help of the following regression equation: R&D Share = a + b 1 t + b 2 D + b 3 D.t Here R&D Share represents, in each of two separate regressions, the figures in the last two columns of Table 2, t is the time period, and D is a dummy variable taking the value of 1 for 1988 (the year Bill C-22 came into effect) and all subsequent years, and zero otherwise. The hypothesis to be tested is that the coefficient b 3 is positive and significantly different from zero, that is, that an upward change in the trend occurred after (The term b 2 D allows for change in the intercept.) Ideally, the test for a change in trend should be based on a fully specified structural model of determinants of R&D. However, estimation of such a model is not feasible, partly because of the insufficient length of consistent R&D time series, and partly because of the lack of availability of information on the many determinants of R&D indicated by economic theory. However, as explained above in the section on Methodology, the dependent variable is formulated in the share form with the specific objective to control for the determinants of R&D spending common to all Canadian industries.

6 34 Bohumír Pazderka TABLE 2 Canada Total Intramural R&D Expenditures by Industry (Millions of Current Canadian Dollars) Year Pharma and Manufacturing All Pharma R&D as % Pharma R&D as % Medicine Industries of Manufacturing of All Industries p i Notes: 1. i describes intentions, and p stands for preliminary. 2. The definition of R&D expenditures reported by Statistics Canada is discussed in detail, for example, in Statistics Canada (1997, pp ). While it closely follows the definition in the Income Tax Act, Statistics Canada also includes certain R&D expenditures (e.g., on land and buildings) which cannot be claimed for income tax purposes under the Income Tax Act. Source: Statistics Canada, Industrial Research and Development, Catalogue No , various years. The estimated coefficients of the simple regressions specified above are reported in Table 3, equations 1 and 2. 6 The coefficients on the slope dummy variable (b 3 ) are positive both when the dependent variable is pharmaceutical R&D as a percentage R&D spending in all Canadian industries, and as a percentage of Canadian manufacturing R&D spending. Both coefficients are statistically significant at the 99-percent level of confidence (the t-values are 5.8 in both equations). The hypothesis that no change in the trend of Canadian pharmaceutical R&D spending occurred at the time Bill C-22 was enacted is rejected in favour of a significant increase in trend. A Comparison of Pharmaceutical R&D in Canada with the OECD As shown in the preceding section, the hypothesis that, starting around the time of enactment of Bill

7 Patent Protection and Pharmaceutical R&D Spending in Canada 35 TABLE 3 Regression Results Test for Change in Trend (maximum likelihood estimates, with adjustment for serial correlation) Eq. Dependent Const. Time Intercept Slope R 2 adj. D.W. Variable Term Dummy Dummy Statistic 1 Pharm. share of Total Can. R&D (6.9) (-0.6) (-5.2) (5.8) 2 Pharm. share of Can. manuf R&D (9.5) (-1.6) (-4.9) (5.8) 3 Can. share of OECD pharm R&D (15.4) (-0.5) (-5.2) (6.2) Real Growth 4 Can. pharm. R&D (5.9) (6.1) (-12.4) (14.9) 5 OECD pharm. R&D (9.0) (8.6) (-3.5) (4.3) Log Real Growth 6 Can. pharm. R&D (47.9) (5.8) (-3.6) (4.3) 7 OECD pharm. R&D (108.9) (12.7) (-0.3) (0.9) Pharm. share of total R&D 8 Canada (22.5) (-4.9) (-7.4) (9.8) 9 OECD (32.9) (4.4) (-5.4) (6.6) Pharm. share of manuf. R&D 10 Canada (16.2) (-1.9) (-8.7) (10.9) 11 OECD (39.6) (6.1) (-8.5) (10.6) Pharm. share of chemical R&D 12 Canada (4.3) (0.6) (-4.9) (5.1) 13 OECD (52.7) (11.4) (-4.5) (5.7) Notes: 1. The figures in parentheses are the t-values. 2. The OLS estimates are very similar and are available from the author.

8 36 Bohumír Pazderka C-22, pharmaceutical R&D grew disproportionately with the R&D in other sectors of the Canadian economy is indicated. The possibility remains, however, that the faster growth is due to factors specific to the pharmaceutical industry, regardless of country. For example, an acceleration of R&D spending could have been caused by a paradigm shift in pharmaceutical research methodology, or the occurrence of a cluster of innovations. If this were the case, pharmaceutical R&D in relation to R&D spending in other industries should be exhibiting a similar pattern in other countries as well. The ANBERD database developed by the OECD (1996, 1997) contains information on Business Expenditures on R&D (BERD), by industry, for 15 OECD countries (Australia, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands, Norway, Spain, Sweden, United Kingdom, United States), for the period In what follows, Canada is compared with the aggregate of these 15 countries (referred to as the OECD ). Whenever necessary, the ANBERD data given in national currencies were converted into US dollars, using the purchasing power parity exchange rates developed by the OECD. The comparisons are presented in the form of plots of the relevant time series and in the form of regression tests with the help of the equation specified above. The first set of tests focuses on pharmaceutical R&D, without reference to other industries. Figure 1, illustrating Pharma R&D Spending (Canada as Percentage of OECD), shows a slow erosion of Canada s share until 1984, and then a relatively steady climb. The corresponding regression estimates are reported as equation 3 in Table 3. The difference in slopes between the pre- and post-1988 periods is statistically significant at the 99-percent level of confidence (the t-value for the slope dummy coefficient b 3 is 6.2). Figure 2 plots the rate of growth of pharmaceutical R&D expenditures in Canada together with the rate of growth in the rest of OECD. The two series track each other very closely until 1989; after that year Canada s rate of growth surges ahead of that of the OECD. The slope dummy coefficient b 3 in the corresponding regressions (equations 4 and 5 in Table 3) is positive and statistically significant at the 99-percent confidence level both for Canada and OECD, indicating a change in trend both for Canada and the rest of the OECD, perhaps because of a change in pharmaceutical research paradigm. A comparison of the magnitude of the estimated coefficient b 3 in regressions 4 and 5 shows that the coefficient for Canada is almost six times larger than the coefficient for the remaining 14 OECD countries. When the dependent variable is in logarithms (equations 6 and 7), the estimated slope dummy coefficient for Canada is again positive and statistically significant at the 99-percent level of confidence, while the coefficient for the rest of the OECD is not statistically significant. It is reasonable to conclude from this analysis that the passage of Bill C-22 had a positive effect on the growth of R&D spending in Canada after The second set of tests uses the ANBERD data to determine whether the pharmaceutical industry in all OECD countries is somehow different from other industries. In addition, it asks whether the pharmaceutical industry in Canada after 1987 is more different than its counterparts in other countries. In Figure 3, pharmaceutical R&D is plotted as a percentage of R&D in all industries. For the OECD as a whole, this ratio shows a steady rise during the whole period; for Canada, however, it declines until about 1987 and rises thereafter. The slope dummy coefficients b 3 in the corresponding regressions 8 and 9 in Table 3 are statistically significant at the 99-percent confidence level both for the OECD and for Canada. The coefficient for Canada, however, is almost twice that for the OECD. When pharmaceutical R&D is plotted as a percentage of manufacturing (Figure 4), and chemical

9 Patent Protection and Pharmaceutical R&D Spending in Canada 37 FIGURE 1 Canadian Pharma R&D as a % of OECD Pharma R&D % Year FIGURE 2 Rate of Growth of Real Pharma R&D Expenditures (1973=100) Canada OECD % Year

10 38 Bohumír Pazderka FIGURE 3 Pharma R&D as a % of All Industries R&D % Canada OECD Year FIGURE 4 Share of Pharma R&D in Manufacturing R&D Canada OECD % Year

11 Patent Protection and Pharmaceutical R&D Spending in Canada 39 FIGURE 5 Share of Pharma R&D in Chemical R&D % Canada OECD Year industry R&D (Figure 5), the rising trend in Canada begins already in 1984 (see endnote 5 for a possible explanation). In regressions where the dependent variable is pharmaceutical R&D as a share of manufacturing R&D (equations 10 and 11 in Table 3), the slope dummy coefficient b 3 for Canada is twice as large as the coefficient for the rest of the OECD. When the dependent variable is pharmaceutical R&D as a percentage of the chemical industry s R&D (equations 12 and 13 in Table 3), the slope dummy coefficient for Canada is almost six times larger. Again, these patterns are consistent with the hypothesis that the enactment of Bill C-22 had a significant positive effect on pharmaceutical R&D spending in Canada. The hypothesis of a structural change was also tested with the help of cumulative sum and cumulative sum of squares of recursive residuals (see e.g., Johnston 1984, pp ; SHAZAM 1997, pp ). The various definitions of the dependent variable listed in Table 3 were regressed on a constant term and time and the cumulative sums were plotted. For example, in the first regression, the dependent variable is based on Statistics Canada data from Table 2, and is defined as the Canadian pharmaceutical industry s share of total Canadian industrial R&D (as in equation 1 in Table 3). The plot, shown in the Appendix, indicates evidence of structural change exceeding the 5-percent significance bound in the year The plots of cumulative residuals from the other equations in Table 3 show similar patterns, with structural change occurring mostly during the period These results are broadly consistent with the hypothesis that the 1987 amendments to the Patent Act had a significant impact on pharmaceutical R&D spending in Canada. Patent Protection and Location of R&D Activity by Multinational Corporations It is conceivable that the apparent increase in pharmaceutical R&D spending in Canada after 1987 can be attributed to the growing attractiveness of Canada

12 40 Bohumír Pazderka as a location for R&D in general, not just in the pharmaceutical industry. This hypothesis is explored with the help of information on the geographic distribution of R&D expenditures by foreign subsidiaries of US-owned corporations. Historically, Canada has been a leading location for the activities of US multinationals, given its geographic, linguistic, and cultural proximity to the United States. As shown in Table 4, in 1966 Canada represented over 30 percent of US foreign direct investments in all industries and in manufacturing, and slightly less in chemicals and allied products. By 1989, Canada s share of investment dropped below 17 percent in all industries and in chemicals, and to about 20 percent in manufacturing. (Disaggregated data for pharmaceuticals are not available.) Canada s share of R&D performed abroad by USowned multinationals declined sharply between 1966 and During the period (when compulsory licensing was in effect), Canada s share of pharmaceutical R&D declined from 8.6 percent to 7.3 percent, while the other industry aggregates except chemicals rose. In contrast, the share of pharmaceutical industry share was higher in the post-compulsory licensing year 1989, while that of the other industry aggregates either declined, or rose marginally (the exact year of change cannot be established). The US Pharmaceutical Manufacturers Association (PMA) performs biannual surveys of its member companies, which include information on the geographic breakdown of their R&D expenditures. While participation in the surveys is voluntary and the number of respondents varies from year to year, this is the only information of its kind available. The data summarized in Table 5 indicates that spending abroad as a percentage of total R&D of PMA member companies changed only marginally since 1987, suggesting the absence of a significant trend toward decentralization of R&D away from headquarters of US-owned multinational pharmaceutical corporations. Canada s share of the total spending abroad TABLE 4 Foreign Investment and R&D Spending of US Multinationals (Canada as percentage of all countries) All Industries Manufacturing Chemicals Pharmaceuticals Year Invest. R&D Invest. R&D Invest. R&D Invest. R&D N.A. N.A N.A N.A N.A Notes: 1. Investment shares are based on U.S. direct investment position abroad on a historical cost basis. 2. R&D spending shares are for majority-owned nonbank affiliates of nonbank U.S. parents. For the years 1966 and 1977, they represent R&D expenditures by affiliates (for themselves and for others). For the later years, they represent R&D expenditures for affiliates (by themselves and by others). Sources: 1. US Department of Commerce, Survey of Current Business for years , and Special Tabulation, Bureau of Economic Analysis, for years (for investment shares). 2. US Department of Commerce, U.S. Direct Investment Abroad. Benchmark Survey Data (for R&D shares).

13 Patent Protection and Pharmaceutical R&D Spending in Canada 41 TABLE 5 R&D Spending of PMA Member Companies (in US dollars) Year Total R&D R&D Abroad Abroad as % Canada as % ($ mill.) ($ mill.) of Total of Abroad N.A N.A * Note: * R&D abroad in 1995 was affected by merger and acquisition activity. Sources: 1. Pharmaceutical Manufacturers Association, Backgrounder: U.S. Pharmaceutical R&D and Sales, February Pharmaceutical Manufacturers Association, Annual Survey Report, various issues; personal communication (Canada ). moved between 4 and 7 percent during the late 1970s and early 1980s, and rose somewhat in the second half of the 1980s. In 1989, that is, one year after Bill C-22 came into effect, it jumped to the historically high share of 10.4 percent, dropped to 7.1 percent in 1992, and rose to 8.5 percent in While these numbers provide some indication that the Canadian pharmaceutical industry in post-1987 may have benefited disproportionately from any decentralization of R&D away from headquarters, this evidence is not very strong. IMPACT OF TAX SUPPORT FOR R&D The principal features of the current federal government system of tax incentives for R&D have been in place since The definition of eligible R&D activities was expanded in 1985, payments by corporations for R&D performed on their behalf by approved organizations, such as the Medical Research Council, became eligible for R&D tax support in 1986, and the carryforward period for R&D tax credit was extended in The processing of claims was simplified in 1988 and further

14 42 Bohumír Pazderka improvements were made in 1992 among them expanded eligibility for capital R&D expenditures (Iqbal 1995). While no major change in the federal incentives occurred as Bill C-22 was enacted, a significant change did occur in the province of Quebec, which operates the most generous of all of the provincial systems of R&D incentives. The tax credit on R&D salaries, introduced in 1983 at 10 percent was raised to 20 percent in 1987 and, for small enterprises only, to 40 percent in The preliminary version of a study prepared for the provincial Conseil de la science et de la technologie (Lebeau 1996) argues that the enhancement of incentives contributed significantly to increased R&D activity in the province. These incentives apply to all industries; the analysis in the previous section, comparing pharmaceuticals with other Canadian industries, therefore remains unaffected. However, if it can be shown that pharmaceutical R&D spending in Quebec after 1987 behaved differently from other provinces, the analysis comparing pharmaceutical R&D in Canada with that in the OECD countries may be contaminated by the change in R&D incentives coinciding with the enactment of Bill C-22. Data on regional distribution of pharmaceutical R&D expenditures in Canada are available from two different sources, and the evidence is somewhat mixed. As shown in Table 6, Statistics Canada data indicate a substantial increase in Quebec s share, starting around 1989; it reached a peak in 1990 and seems to have stabilized at a somewhat lower level in subsequent years. The PMPRB data series starts only in 1988, and Quebec s share of Canada s R&D expenditures remained relatively stable throughout the whole period. 8 The data on provincial distribution of pharmaceutical R&D expenditures do not easily lend themselves to econometric analysis, mainly because of TABLE 6 Quebec s Share of Canadian Pharmaceutical R&D Expenditures (percentages) Quebec s Share Quebec s Share Year StatsCan Data Year StatsCan Data PMPRB Data N.A N.A Note: Numbers in the StatsCan column are calculated from data on total intramural R&D expenditures, reported by Statistics Canada. Numbers in the PMPRB Data column represent current R&D expenditures by location, reported by the Patented Medicine Prices Review Board. Sources: 1. Statistics Canada, Industrial Research and Development, Catalogue No , various years. 2. Patented Medicine Prices Review Board, Annual Reports, various years.

15 Patent Protection and Pharmaceutical R&D Spending in Canada 43 the small number of available observations. As reported by Statistics Canada (Catalogue No ), until 1984 Quebec and Ontario jointly represented virtually 100 percent of Canadian pharmaceutical R&D and even in 1995 their combined share was 95.9 percent. The impact of the Quebec tax change was tested with the help of the following regression on 34 pooled, cross-sections of time-series observations (provincial shares calculated from this data are reported in Table 6) for the two provinces over the period : Dep. var. = a + b 1 t + b 2 D + b 3 D.t + b 4 DO + b 5 DQ + b 6 DQ.t where the dependent variable is defined either as provincial share of Canadian pharmaceutical R&D expenditures for Ontario and Quebec, respectively, or as the real dollar value of provincial expenditures; D is a dummy variable taking the value of 1 in 1988 and all subsequent years, and zero otherwise, DO is a dummy variable that takes a value of 1 for Ontario observations and zero otherwise; and DQ is a dummy variable taking the value of 1 when the observations apply to Quebec in the year 1988 and all subsequent years (i.e., when the Province of Quebec R&D tax credits were in effect), and zero otherwise. The results of the cross-section of time-series estimation (SHAZAM 1997, ch. 23) are mixed. When the dependent variable was defined as provincial share, none of the estimated coefficients were statistically significant at the 99-percent level of confidence. When the dependent variable was the real provincial R&D spending, coefficient b 3 was positive (t-value 9.6), b 4 was negative (t-value 3.4), and b 5 was positive (t-value 5.0), while b 6 was not statistically significant (t-value 0.2). When the dependent variable was the logarithm of real provincial spending, b 3 was not significant at the 99-percent confidence level (t-value 1.4), b 4 was negative (t-value 4.8), but b 5 was not statistically significant (t-value 0.3), nor was b 6 (t-value 0.6). These results are largely consistent with the hypothesis of a structural change after However, this test suggests rejection of the hypothesis that the change in Quebec tax law had a significant independent impact on pharmaceutical R&D spending in the province. CONCLUSIONS A number of indicators of pharmaceutical R&D spending in Canada show an increase in R&D activity starting around the year 1987 (the time of enactment of Bill C-22). The objective of this paper was to determine whether the observed increase in R&D spending can be attributed to Bill C-22. The data presented in Table 1 illustrate a dramatic acceleration in R&D spending of the PMAC member companies after 1988, but a slowdown in the rate of growth in the period. The PMAC commitment to raise the R&D-to-sales ratio to 10 percent by 1996 was satisfied already in The growth in real R&D spending of all Canadian pharmaceutical companies (calculated from Statistics Canada data reported in Table 2) also shows a similar pattern. During the ten years between 1975 and 1987, the real R&D spending in the pharmaceutical industry almost exactly doubled, as did R&D spending in Canadian manufacturing, while R&D spending in all industries increased almost 2.9 times. During the ten years between 1988 and 1997, pharmaceutical R&D spending in Canada increased 3.4 times, while R&D spending in manufacturing increased only 1.4 times, and R&D spending in all industries almost 1.6 times. The results of regression analysis performed both with Canadian and OECD data are also consistent with the hypothesis of a structural change taking place after While the preponderance of data analysis and statistical tests presented in this paper support the conclusion that the tightening of patent protection in Bill C-22 led to a change in trend in pharmaceutical R&D spending in Canada, this increase should not be attributed exclusively to the economics of patent protection. It is important to note that during the debate on the merits of Bill C-22 the PMAC member

16 44 Bohumír Pazderka companies made an extensively publicized commitment to raise the ratio of R&D to sales from less than 5 percent prevailing in 1984 to 8 percent in 1991, and 10 percent in 1996 (PMAC 1992, p. 16). The promise to increase R&D spending was, of course, conditional on the tightening of patent protection and vice versa; hence the contributions of the two cannot be separately identified. It seems reasonable to conclude, however, that the increased patent protection alone, without the accompanying public commitment by the PMAC, may not have resulted in an R&D spending increase of the magnitude observed in Canada. NOTES This paper draws upon a larger study prepared for the OECD international research project on Biotechnology and Medical Innovation: Socio-Economic Assessment of the Technology, the Potential and the Products. An earlier version of the paper was presented at the annual meeting of the Canadian Economics Association in St. John s, Newfoundland. I wish to thank the editor and three anonymous referees for very useful comments. Shawn Rosemarin provided excellent research assistance. 1 Since 1923, the Patent Act has permitted compulsory licensing of pharmaceuticals to manufacture active ingredients in Canada. Between 1923 and 1969, only 22 compulsory licences were granted, primarily because manufacturing in Canada was not profitable due to the small size of the market. Several commissions of inquiry during the 1960s concluded that drug prices in Canada were too high. As a remedy, amendments to the Patent Act enacted in 1969 allowed for compulsory licensing of imports of patented drugs or their active ingredients. Between 1969 and 1987, there were 765 applications filed, mostly by Canadian-owned generic companies. About 400 licences were granted, almost all of them to import active ingredients (Heller 1995, p. 222). Early in the process, the courts established the royalty rate at 4 percent of the net selling price of the drug in final dosage form. In order to maximize the benefit from the increased price competition, provincial governments passed amendments to their pharmacy acts, mandating substitution of a prescribed drug by the cheapest generic equivalent in the pharmacist s inventory. 2 Between February 1988 and August 1995, 16 large mergers took place in the international pharmaceutical industry. Some of them led to rationalization of Canadian operations, including R&D, and to job losses (CDMA 1995). The exact impact on pharmaceutical R&D spending in Canada is unknown. 3 Drug companies which do not hold patents and thus do not report to the PMPRB have also increased their R&D spending. For example, the average R&D-to-sales ratio for the member companies of the Canadian Drug Manufacturers Association in 1993 was reported to be 13 percent, up from 7 percent in 1988 (CDMA 1994a). It is hard to avoid the inference that the increased R&D spending of CDMA member companies was stimulated by the stronger patent protection, despite the association s continuing opposition to Bill C-22 and C The PMPRB (1989, pp ) defines applied research to include creating new or improved products or processes mainly through research in chemistry, pharmaceutical dosage forms, studies in animals (pre-clinical research) or studies in humans (clinical research). 5 For example, the last two columns in Table 2 record a dip in the year 1984 and a peak in the year A potential explanation of the 1984 dip may run as follows: the Statistics Canada annual reports on Industrial Research and Development (Catalogue No ) for years 1985 and 1986 include a footnote that improved coverage and response have increased observed R&D expenditures by about 10 percent for these years. A possible inference is that the coverage for 1984 was poor, perhaps more so in the pharmaceutical industry than in others. The 1990 peak could be due to a temporary surge in R&D spending in Canada by foreign-owned multinationals (see also Table 5) which has since returned to more traditional levels. The 1984 and 1990 deviations from the trend are also visible in the graphs for Canada in Figures 1 to 5 below (based on the OECD data). 6 The coefficients reported in Table 3 are maximum likelihood estimates with adjustment for serial correlation (SHAZAM 1997). OLS results are very similar and are available on request from the author. 7 The OECD reports on business enterprise expenditures on R&D (BERD) in two formats: first, the official database (OFFBERD) and second, the analytical database (ANBERD). The international comparisons in this paper are based on the ANBERD database which was

17 Patent Protection and Pharmaceutical R&D Spending in Canada 45 created by the OECD in collaboration with the national statistical agencies to overcome the problems of international comparability and time discontinuity associated with the OFFBERD data, reported to the OECD by the member countries (OECD 1996, p. 16, and ch. II). 8 The discrepancy in the behaviour of the two data series can be explained by differences in the industry coverage by the two agencies. The PMPRB reports only R&D spending by patent-holding companies, while Statistics Canada covers all companies. The post-1989 rise in Quebec s share of the total Canadian pharmaceutical R&D spending may reflect an increase in R&D spending by small start-up biotechnology companies in Quebec, which did not yet have a patented product on the market, and eagerly responded to the improved patent protection and tax incentives. An additional factor may be the difference between the PMPRB and Statistics Canada definitions of R&D: while the former reports R&D expenditures eligible for tax credit under the provision of the Income Tax Act in effect on 1 December 1987, the latter reflects the tax legislation in effect in the current year and includes additional R&D expenditures, such as those on land and buildings. However, these definitions apply across the country and across all industries and should not affect the R&D spending share of any particular industry in any particular province. REFERENCES Boards of Trade (1996), A Prescription for Success, a joint position paper on the pharmaceutical industry prepared by the Boards of Trade of Metropolitan Toronto and Metropolitan Montreal (Toronto: Board of Trade of Metropolitan Toronto). Canadian Drug Manufacturers Association (CDMA) (1994a), Quick Facts on Research and Development (Toronto: CDMA). (1994b), Submission to the Standing Committee on Foreign Affairs and International Trade Regarding Bill C-57 (An Act to Implement the Agreement Establishing the World Trade Organization) (Toronto: CDMA). (1995), The Brand-Name Pharmaceutical Industry Canadian Job Losses and Plant Closings (Toronto: CDMA). Department of Foreign Affairs and International Trade (DFAIT) (1996), The Case for Investing in Canada (Ottawa: DFAIT). Eastman, H. C. (1985), The Report of the Commission of Inquiry on the Pharmaceutical Industry (Ottawa: Supply and Services Canada). Gambardella, A. (1995), Science and Innovation. The U.S. Pharmaceutical Industry During the 1980s (Cambridge: Cambridge University Press). Generics Want Relief, (1996), The Globe and Mail, 20 May, p. B1. Grandstrand, O., L Håkanson and S. Sjölander (1993), Internationalization of R&D: A Survey of Some Recent Research, Research Policy 22: Heller, J.G. (1995), Background Economic Study of the Canadian Biotechnology Industry, study submitted to Environment Canada, Health Canada, and Industry Canada (Toronto: James G. Heller Consulting). Iqbal, M. (1995), R&D Tax Incentive Comparisons: Canadian and U.S. Large Manufacturing Industries, Canadian Tax Journal 43: Johnston, J. (1984), Econometric Methods, 3d ed. (New York: McGraw-Hill). Lebeau, D. (1996), Mesures fiscales d aide à la R-D (Québec: Conseil de la science et de la technologie). Organization for Economic Cooperation and Development (OECD) (1996), Research and Development Expenditures in Industry, , 1996 ed. (Paris: OECD). (1997), Research and Development Expenditures in Industry, ed. (Paris: OECD). Palda, K.S. (1993), Innovation Policy and Canada s Competitiveness (Vancouver: The Fraser Institute). Pharmaceutical Manufacturers Association of Canada (PMAC) (1992), A Stronger Patent Act, The Hill Times, Advertising Supplement, 19 November. (1993), Towards Canadian Pharmaceutical Patent Law Reform, a submission to the Senate Committee on Banking, Trade and Commerce Studying Bill C-91 (Ottawa: PMAC). (1995a), Promises Made... Promises Surpassed! A Special Report (Ottawa: PMAC). (1995b), The Annual Review (Ottawa: PMAC). Patented Medicine Prices Review Board (PMPRB) (1989), First Annual Report 1989 (Ottawa: PMPRB). (1996), Eight Annual Report for the Year Ended December 31, 1995 (Ottawa: PMPRB). (1998), Tenth Annual Report for the Year Ended December 31, 1997 (Ottawa: PMPRB). Re$earch Money (1996), Canada s Top-100 Industrial R&D Spenders Maintain Record Growth Rate for Second Year Running, Re$earch Money, 17 July (Ottawa: Evert Communications).

18 46 Bohumír Pazderka SHAZAM (1997), User s Reference Manual Version 8.0 (New York: McGraw-Hill). Statistics Canada (1997), Industrial Research and Development Intentions (Ottawa: Statistics Canada). YORKTON (1995), Canada s Generic Giants Diversify into Biotechnology, Health Sciences for Investors, 8 August (Toronto: Yorkton Institutional Group). APPENDIX FIGURE A1 Cumulative Sum of Residuals (CUMSUM) from Equation 1 in Table CUSUM BOUND YEAR