WPCB 2 B9P Z Courier 10cpiCourier 10cpiCourier 10cpi (Bold)Line Printer 16.67cpiHP LaserJet IIIHPLASIII.PRSx  @,\,A]X@2<#LX#3xCourier 10cpiCourier 10cpi (Bold)HPLIIIAD.PRSx  @,\,ISX@2xxx,x  @87X@2xxx,x  `B7Xtk#HHH,#H H@7@>|Z > Z Z Z >>R >>>>| >>>> b R  b 2Y07oCourier 10cpiPres Elite 16.67cpi (25inOne!)Courier 10cpi (Bold)Courier 10cpi (Italic)05-14-96 11:07a  2@ 2  D     =Њ! Longer Patents For Increased Generic Competition:  2  @ The WaxmanHatch Act After One Decade ă Henry Grabowski and John Vernon Forththcoming in PharmacoEconomicsp=DDԌ   D pLonger Patents For Increased Generic Competition: @ The WaxmanHatch Act After One Decade  2 X01Í ÍX1*Í Í Henry Grabowski and John Vernon*  2@ ԍThis paper is to be presented at the Tufts University Center for the Study of Drug Development Conference in Talloires, France, July 21, 1995. Please do not quote without permission of the authors.փ X1*Í ÍX01Í Í h#Duke University &June 1995 The 1984 Drug Price Competition and Patent Term Restoration Act (the WaxmanHatch Act) is unique in many ways. This Law was the first change in U.S. patent terms since 1861. It represented a delicate political compromise. It simultaneously lowered the barrier to entry for generic drug firms and increased patent terms for new drugs delayed by FDA regulation. According to Senator Hatch, in his speech introducing the legislation in the Senate, "The public receives the best of both worldscheaper drugs today and better drugs tomorrow." The WaxmanHatch Act became effective on September 24, 1984. We now have a decade of experience with it. In this paper, we analyze how the Act has performed in practice. The next section examines how generic competition has changed under this law. In the following section, we then consider how it has affected patent lives for new drug introductions since 1984. The final section considers how the Act has impacted consumers and pharmaceutical innovation. Specifically, we wish to ascertain whether Senator`"0*((@@ Hatch's forecast that the Law would produce both cheaper drugs and better drugs has, in fact, been the case.  I. Generic Entry and Price Competitionă Title I of the WaxmanHatch Act established an Abbreviated New Drug Application (ANDA) procedure for generic drugs. Under this ANDA procedure, generic drugs need only demonstrate bioequivalence to the pioneer's brand. Prior to this Law, generic drugs could not rely on the safety and efficacy submitted by the branded product and had to duplicate many of pioneer's brand tests to gain FDA  2 approval.*@! 20 ԍBefore the 1984 law, generic firms could cite evidence on safety and efficacy published in the scientific literature for post62 drugs in lieu of performing these studies. However, since a considerable amount of the evidence needed to obtain FDA approval was not part of any published work, many studies had to be repeated to gain approval. This was a costly procedure and consequently, many of these drugs had little or no generic competition even several years after patent expiration (Grabowski and Vernon, 1986).* In an earlier article, we examined the effect of generic entry on pricing behavior and market shares for the period 19841987 (Grabowski and Vernon, 1992). That study covered 18 drugs whose patents expired in the period immediately following the passage of the 1984 Drug Price Competition and Patent Term Restoration Act. In this section, we extend our analysis to 22 drugs that experienced generic competition for the first time in the period 19891993. As we observed in our earlier study, prior to the 1984 Act there were a number of factors that led to strong brand loyalty for the brand name drug even after patent expiration. For example, 0*((@@ physicians generally gained experience with brand name drugs during the patent exclusivity period, and upon patent expiration they continued to prescribe the brand name drug. Often they had little incentive to prescribe the lower cost generic drug and many doctors were uninformed about the relative prices. Furthermore, state antisubstitution laws instituted in the 1950s and 1960s prohibited pharmacists from substituting the cheaper generic drugs. Beginning in the mid1970s, however, there have been a number of institutional changes in pharmaceuticals that have led to greater price sensitivity. First, the state antisubstitution laws have all been repealed. The new substitution laws permit and often require pharmacists to substitute lower cost genericsunless the physician specifically requires that the brand product be dispensed. (As we will document later, pharmacists generally have a profit incentive to dispense the generic product rather than the brand product.) In addition, thirdparty payers such as Medicaid instituted requirements limiting reimbursements to generic price levels. The growth of managed care programs, pharmacy benefit managers (PBMs) and health maintenance organizations (HMOs) have significantly increased the incentives for generic utilization in the last several years. A recent analysis estimated that over 115 million beneficiaries were covered by employerfunded PBM services in 1994, and incentives for generic usage is one of the first steps taken under these plans (Mandelker, 1994). These developments make%0*((@@ an analysis of recent patent expirations especially interesting at this time. In this section, after explaining the nature of our data sample for the 19891993 period, we will contrast our new findings with our earlier 19841987 results. That is, our 19841987 results found that generic products entered at a significant price discount to the brand product and, contrary to the period before the 1984 Act, won close to half the market within two years. It was also the case that the brand product did not compete on a price basis with the generics. Hence, our first analysis will be a broad comparison of the 19891993 period with the 19841987 period. Next, we will present a more detailed and disaggregated analysis of the recent period. The final topic in this section of the paper is the nature of price competition at the pharmacy level. That is, given that generic prices fall to less than half of the brand price, to what extent do these price cuts flow through to the final consumer? The 19891993 Data Sample We assembled data for a sample of 22 brand products first  2x exposed to generic competition in the period 19891993.ox! 2 ԍTable A1 in the appendix lists the drugs with their date of generic entry.o Our objective was to include all of the major products; however, we also include for some of our analysis four products that barely failed to meet our criterion of at least $50 million in sales at(# 0*((@@ the time of patent expiration. Hence, we have a sample of 18 major drugs and four smaller drugs. For each product, we identified the most popular dosage sizes  2 from the data audits of IMS America Inc.d! 2 ԍIMS America Inc. U.S. Drug Store and Hospital Sales (19891993).d These data sources also provide an estimate of total sales of each dosage size by manufacturers and wholesalers to drugstore and hospital outlets. Sales are available in both dollars and physical units. Using these data, we computed prices at various time intervals for the brand products and the generic entrants. Hence, the prices represent the average cost per unit paid by drugstores and hospitals for the most popular dosage size for each product. Market shares in units for the brand and generics were also computed at the same points in time after entry. Comparison of the 19841987 and 19891993 Period Table 1 provides a summary of our findings for the entire period, 19841993. As noted above, results for the 19841987 period were reported in our 1992 article, and covered 18 major brand name drugs that first experienced generic competition during those years. Here, we report those findings for comparison with our new results for a later period. For more meaningful comparison, we have broken the earlier study's results into two  2! subperiods: 19841985 and 19861987.! ! 2H& ԍOne of the ten drugs in the 19841985 category actually experienced generic competition in mid 1983. For the more recent period,!x0*((@@ in order to have two full years of data after generic entry, we are limited to seven drugs for the years 19891991. The first row of Table 1 shows that the average brand name price index in the first year after generic entry rose 6 percent for 198485 drugs, 8 percent for 198687 drugs, and 6 percent for 198991 drugs. Hence, the pattern of the brand name product not dropping its price to compete with the much lower priced generics continues in the more recent period. As the table shows, similar small increases in the brand name price index hold for the second year after entry. In contrast to the increases of the brand name price index, the second row shows that quite similar large decreases in the generic price index took place over the two year period. The three periods all experienced price decreases of about 35% by generics over the two year period. The third row also evidences similar patterns for the ratio of generic to brand prices. That is, at entry the generic price is about 60% of the brand price and falls to a bit over one/third of the brand price after two years. The final row of Table 1, in contrast to the results described above, does seem to indicate differences among the three periods. The final row shows how the average generic market share after two years has increased over time. In the first period (198485) the generics averaged 45% of the market in units. In the second period (198687) the generic share rose to 54%, and in the third (198991) it rose even higher to 59%. (Figure 1 shows these results graphically.) This finding is consistent with the general'0*((@@ observation that managed care and other demand side pressures are increasing generic utilization over time. As we shall see, by disaggregating the recent period and looking at shorter periods after generic entry, we found that the extent of generic utilization has intensified significantly within the 198993 period.  The Recent Period: 19891993 We focus now on the recent data period, 19891993, for which we have data on 18 major drugs and four small drugs. Table 2 contains results similar to those in Table 1, but for the 18 major drugs of the recent period. The drugs are broken down into three categories: six drugs that had generic competition initially in 198990, five drugs that had generic competition in 199192, and seven drugs that experienced generic entry in 1993. Table 2 and Figure 2 suggest that the trend of increasing generic utilization is continuing rather dramatically over time. For example, the average generic market share eighteen months after entry increased from 47% for the 198990 period to 72% for the 199192 period. The average ratio of generic to brand name price fell from 50% to 42% for this same time interval. If we focus on a much shorter periodsix months after entry we observe that generic utilization increases from 31% to 44% to 51% over the three periods. The last period, the year 1993, is so recent that we can only measure shares six months after entry. The year 1993 may have been unusual in that two brand name suppliers tried new strategies to thwart generic competition. An'0*((@@  2 articlex! 2X ԍ"The Drugmakers vs. The Trustbusters," Business Week, September 5, 1994, p.67.x in Business Week indicates that Upjohn and Syntex, the suppliers of the 1993 drugs, Xanax and Naprosyn, introduced their own generic version of their brand name product in October 1993. By June 1994, according to the article, Upjohn itself had 90% of the generic market and Syntex had 68%. (Our own data only six   months post generic entry, show total generic shares to be 68% and 77%, respectively.) They obtained these large shares by introducing their own generic versions before the patents expired. While this strategy gave them an advantage in being first on the  2 marketx ! 2 ԍBeing first is an advantage for several reasons. For example, by getting drug chains to stock the first generic, pharmacists and customers become acquainted with that particular pill shape and color and will not want to change. Also, the initial generic company is usually given the opportunity to simply match a subsequent generic's price and thereby maintain the pharmacies' accounts., it did not slow down the generic price cuts. The same article notes that the generic price of Naprosyn fell to 10% of the brand name's price. Whether such aggressive behavior by the brand name firms is profitable, and will continue in the future, is an open question at this time. ( Continuing to focus on the generic market shares six months post entry, we now consider a descriptive regression in which we also include four additional drugs that failed to meet the $50 million sales criterion. The four small drugs had sales in the year before entry between $23 million and $43 million. Recall the other 18 drugs had minimum sales of at least $50 million. ` 0*((@@ԌThe dependent variable is taken to be generic market share six months post entry, or GENSH. One obvious explanatory variable is, of course, the date of first generic entry. Hence, we define T=1 to be June 1989, and let T represent the number of months from that date. A second explanatory variable is SIZE, or the total sales of the brand name product in the year before entry (in billions of dollars). The hypothesis is that the larger the SIZE, the larger the number of generic entrants will be because of perceived profitability. And, the greater the number of generic suppliers, the more vigorous we expect price competition to be, and therefore  2 the larger the generic market shares.X! 2 ԍA recent econometric study of generic competition shows a table in which price reductions rise with the number of generic suppliers (Caves, Whinston and Hurwitz, 1991). In addition to the explanatory variables described above, we also introduced three dummy variables to pick up the effects of the therapeutic class of the drugs. The dummy variables are ANIF for the antiinflammatory/muscle relaxant group of drugs, CARD for heart drugs, and PSYCH for psychotropic drugs. The sample includes eight heart drugs, seven antiinflammatory/muscle relaxant drugs, three psychotropic drugs, and the remaining four drugs are each in a different class. As noted, the sample is 22 drugs introduced over the period 1989 to 1993. The variable SIZE ranges from a low of $23 million to a high of $614 million. The mean value of SIZE is $208 million.`" 0*((@@ԌThe estimated equation is: GENSH = 0.09 + 0.004T + 0.58SIZE + 0.04PSYCH + 0.002CARD + 0.17ANIF (1.83) (3.30) (4.64) (0.56) (0.04) (3.23)   2x where the adjusted R2 = 0.77, n = 22, and the tstatistics are in parentheses. The statistical results are interesting. As expected, the coefficients of the two explanatory variables, T and SIZE, are both positive and both are statistically significant. Of the three therapeutic class variables, ANIF is the only statistically significant variable. Hence, the results indicate that antiinflammatory/muscle relaxant drugs lose some 17 percentage points more to generics than do other types of drugs (i.e., measured six months after initial generic entry). The regression explains over three quarters of the variation in generic shares across the 22 products. Consistent with the results in Table 2, the coefficient of T indicates that drugs exposed to generic competition at more recent points in time experience greater market losses to generics. This "vintage" effect is approximately 5 percentage points per year. Hence, a drug first experiencing generic competition in June 1993 lost 20 percentage points more of its unit market share after six months compared to one first experiencing generic competition in June 1989, other things equal. Also, the coefficient of SIZE indicates that a larger selling drug by $100 million will lead to a higher generic share of about 6 percentage points.H& 0*((@@ԌFinally, we consider the pricing pattern and market shares for the longest period permitted by our datathree and onehalf years. In particular, the results for the average of the seven major drugs that experienced generic entry for the first time in 198990 are shown in Figure 3. Fourteen quarters after entry, the average generic market share is 71% and the ratio of generic to brand price is 34%. As can be seen, the price ratio has only a very slow decline after about quarter 8 when it reached 41%. On the other hand, the generic market share continues to increase throughout the period. Generic share in quarter 8 is 53% and, as noted above, rises to 71% in quarter 14. Price Reduction Benefits to Consumers As discussed earlier, the 1984 Act was designed to benefit consumers by encouraging generic competition upon patent expiration. Clearly, as Tables 1 and 2 indicate, generic prices are generally much lower than brand prices. For example, Table 2 shows that for 1993 generic entry cases, the generic price six months after entry is 46% of the brand price, or a "price cut" of 54%. In a competitive retail distribution sector, one would expect price cuts at the wholesale level to flow through to consumers. However, the degree of competition in the retail market for prescription drugs is open to question. Among other factors, the amount of price and quality information available to consumers can be relatively costly to obtain. This leads to the expectation that drug stores possess some market power in their pricing of prescription drugs.' 0*((@@ԌBasically, we seek to examine whether pharmacists earn a higher profit per unit on the lower price generic drugs, thereby appropriating some of the potential gains to consumers from lower manufacturers' prices. This question has been investigated for time periods predating the 1984 Act by Masson and Steiner (1985) and Bloom, Wierz and Pauly (1986). Both studies found that drug store margins on generic products are higher on both a percentage and absolute basis. However, since their analysis was undertaken, there has been a dramatic increase in the growth of generic drugs as a result of the 1984 Act and other factors. It is therefore useful to examine this issue for the post 1984 Act period. In order to examine prices at the retail level, we assembled data on the total number of prescriptions, average prescription size, and dollars spent by consumers for the drugs in our sample. These data are collected in the IMS National Prescription Audit. Using the most popular dosage prescriptions, we computed prices on a per pill basis. Because of data availability problems, we were able to obtain complete data on only 15 of the 18 drugs in our 19841987 sample. Similarly, using the IMS Drug Stores Audit, we obtained matching data for the wholesale prices paid by pharmacists on a per pill basis. Table 3 provides a summary of this analysis. (Detailed data on each drug are given in Table A2.) As indicated, all prices are for one year after generic entry. Table 3 shows that the average generic margin of 13.24 cents exceeds the average pioneer margin of 9.86 cents. This is true despite the fact that the average pioneer' 0*((@@ wholesale price is 33.38 cents compared to only 14.85 cents for the generics. Hence, the average percentage markup for generics is 89% compared to 30% for pioneers. These results are consistent with Masson and Steiner (1985) who found that the dollar margin on generics was higher than on pioneers for 23 of 37 drug entities. We found that for 13 of the 15 drugs, the absolute generic margin exceeds the absolute pioneer margin. Masson and Steiner also argued that there tends to be a greater overstatement of generic wholesale prices than of pioneer prices in the IMS data. If true, this implies generics are relatively even more profitable. Now let us consider the issue of interpreting the much lower generic prices as a benefit to consumers. For example, using the average in Table 3, the ratio of the generic price to the brand price at wholesale is 14.85/33.38, or 0.44. Hence, one might incorrectly jump to the conclusion that the 1984 Act has given consumers generic prices that are 56% lower than brand prices. However, the ratio of the average generic price to brand price at the retail level is 28.09/43.24, or 0.65. The true price reduction to consumers of generics over brands is then only 35% (not 56%). Of course, the point here is simply that the retail markup of wholesale prices is not a proportional one. Rather, as noted above, the markup of the brand is 30% and generic markup is 89%. While we do not argue that the same percentage markup should apply, a conservative approach would be to assume that in a competitive retail sector the absolute margin of the brand product could be viewed as a minimum absolute margin of the generic. That is, if'0*((@@ there are pharmacy costs that are independent of the cost of the pillsas there obviously arethen it would seem reasonable to take the absolute brand margin (9.86 cents) as a reasonable floor  2 for the generic (rather than the actual margin of 13.24 cents).V! 2 ԍWe term this "conservative" because it seems reasonable to assume that there must be at least some pharmacy costs that are positively related to the wholesale pricee.g., inventory costs. If so, then the generic absolute margin in a competitive retail sector would be less than the brand absolute margin.V If this is done, then the generic retail price would be 14.85 + 9.86, or 24.71 cents. Then, the ratio of generic price to brand price at the retail level becomes 24.71/43.24, or 0.57. The true price reduction of generics over brands in this hypothetical situation would then be 43%. Summing up, it is incorrect to think of the manufacturers' relative prices as a way of measuring benefits to consumers. In the example above, focusing on the manufacturers' relative prices would suggest that consumers get a 56% "price cut" as a result of the increased generic competition stimulated by the 1984 Act. If one looks instead at actual retail relative prices, the "price cut" would be 35%a significant benefit, but far lower than 56%. Finally, if one makes the conservative assumption that generic absolute margins should ideally at least be no higher than brand absolute margins, then the "price cut" could be 43% in such a hypothetical scenario. The analysis presented here must be qualified by the fact that the IMS data sources do not include all of the discounts to pharmacy retailers by wholesalers and also exclude rebates by the`"x0*((@@ manufacturers to managed care institutions. Indeed, the results in this section are relevant primarily to the cash outofpocket consumer market. The managed care market would need to be studied separately using alternative data sources. Of course, as noted earlier, one problem from a policy perspective is the cost of information on quality and price to consumers. It is not clear what policy measures could be taken to  2 improve this situation. X ! 2H ԍSome states' substitution laws initially required 100 percent pass through of savings to consumers, but this deterred generic dispensing (Grabowski and Vernon, 1979). It is worth knowing, however, that all potential savings created by the vigorous generic competition at the manufacturer level do not seem to be passed through to consumers. II. Patent Restorationă In Title II of the WaxmanHatch Act, Congress sought to reverse the decline in effective patent life that had been occurring over the prior two decades. Both clinical testing and FDA regulatory approval times steadily lengthened after 1962, when  2 the KefauverHarris Amendments were enacted.q ! 2@ ԍFor a survey of this literature, see Grabowski and Vernon, 1983, Ch. 2 and 3.q Because drug firms apply for patents early in the development period, effective patent life by the early 1980s was significantly less than the nominal  2x patent life of 17 years. x@! 2H& ԍEffective patent life in this analysis is measured from the date of FDA marketing approval to the date of patent expiration.x 0*((@@ԌFigure 4 shows the average effective patent life for U.S. new chemical entities (NCEs) over the period 1970 through September 24,  2  1984. X ! 2x ԍFor an analysis of effective patent life in the pre84 period, see Eisman and Wardell (1981) and Kaitin and Trimble (1987). While there are clearly annual fluctuations in this time series, there is also a strong downward trend over time. This is reflected by the fact that the average effective life for the three successive periods 197074, 197579, and 198084 was 12.4 years, 10.0 years and 8.1 years, respectively. Hence, the average effective patent life in the period just prior to the enactment of the 1984 law was roughly half the nominal life of 17 years. If Congress had just instituted ANDAs under the 1984 Act, without also addressing the reductions that had occurred in effective patent life due to increased regulation and other factors, this would have had significant adverse consequences for drug innovation incentives. Title II of the Act was designed to restore a portion of the patent life lost during the clinical testing and regulatory approval periods. In particular, the Act provides, for new drugs, an extension in effective patent life equal to the sum of the NDA (New Drug Application) review time plus onehalf the IND (Investigational New Drug) clinical testing time. There is a maximum extension of five years, and extensions are also capped at 14 years of effective patent life. The law also provides for a 5year data exclusivity period in which no ANDAs can be granted for an NCE. This is designed to provide a basic 5year protection period for NCEs whose patents had# 0*((@@ already expired, and those with very short terms after patent restoration. This data exclusivity doesn't absolutely bar a generic competitor from entering the market. However, during this 5year period, generics cannot rely on a pioneer's data, but have to develop their own safety and efficacy data to receive FDA approval. This is a very high cost barrier that would preclude entry in the first five years of market life, except for very unusual situations. The law also had some transitional features. In particular, investigational drugs in clinical testing on September 24, 1984 were eligible for a maximum patent term extension of two years. Hence, the full benefits of the Act were confined to drugs in the discovery or preclinical testing phases at the time of the law's passage. The Act also granted retrospectively an ANDA exclusivity period of 10 years to drugs approved for marketing in the period January 1, 1982 through September 23, 1984. Another statute passed in the same period as the WaxmanHatch Act was the 1983 Orphan Drug Act. This Act provided a variety of incentives for drugs directed to "orphan" indications or rare diseases, subsequently defined in the law as disease indications with less than 200,000 U.S. patients. These incentives included a sevenyear exclusivity period for orphan drugs. In the analysis which follows, we exclude orphan drugs for which no other patent protection exists other than this 7year exclusivity. We do so# 0*((@@ since our primary interest in this paper is in patent term  2 extensions associated with the WaxmanHatch Act. ! 2 ԍInclusion of this set of orphan drugs in our analysis would yield a downward bias over time to our effective patent life series. This is because the number of orphan drugs has increased over time, and 7year exclusivity term is shorter than the average effective patent life for other NCEs. For an analysis of the Orphan Drug Act, see Shulman et al. (1992).  WaxmanHatch Patent Extensions In computing effective patent life for a specific NCE, we look first to see if a compound patent exists. This is the strongest form of patent protection in pharmaceuticals. Compound patents are also typically the patents chosen for extension by drug firms (only one patent per NCE is eligible for extension under the 1984 Act). If no compound patent exists, we alternatively use an NCE's composition or use patents to compute effective patent life. Finally, if the effective life is less than 5 years from all available compound, composition or use patents, we use the 5year data exclusivity term provided under the Act for all post84 NCEs to compute effective life. Figure 5 shows the average annual effective patent life for NCEs approved for marketing between September 24, 1984 and December 31, 1993 (i.e., the post WaxmanHatch period). As shown in Figure 5, the average effective patent life, including the extension, varies from 9.0 years in 1989 to over 12 years in 1992. The average extension due to the WaxmanHatch Act, also shown in this Figure has been growing over time. It was in excess of 3 years in 1993.!@ 0*((@@ԌThe period since 1991 is the most relevant one to judge the benefits of the 1984 Act. Prior to 1990, there are only a handful of NCEs that were eligible to receive patent extensions in excess  2 of 2 years (i.e., the transition cap).X! 2 ԍThere were, however, some NCEs with little or no effective patent life that received the ANDA exclusivity protection of five years. Starting in 1991, however, there are an increasing number of drugs in each year that are eligible for the full benefits of the Act. This point is illustrated in Figure 6. This graph shows convergence over time between the actual effective patent life and a hypothetical effective patent life curve. This hypothetical curve is computed under the assumption that all NCEs in each year receive the full benefits of the Act. By 1993, all but two drugs in that year's cohort received the full patent restoration benefits. The 1991 to 1993 cohort has an average effective patent life of 11.7 years. This is probably representative of the average  2 patent lives of NCEs currently coming onto the market.@! 2X ԍHowever, it should be emphasized that there is also considerable variability around this average effective life. For example, there are 31 NCEs in the 199193 cohort (44%) that have average effective patent life equal to or above the 14year upper bound on WaxmanHatch extensions, and conversely, there are 19 drugs(27%) with patent lives below 10 years (excluding those with only orphan drug protection). There also were 11 orphan drugs with 7year exclusivity as their main property rights protection. If Congress had not included the patent restoration part of the 1984 Act, the representative NCE in this cohort would have had an average effective patent life of 9.3 years. Hence, the 1984 Act 0*((@@ has produced a significant increase in patent protection periods for new drug introductions.  GATT Extensions One interesting recent development for pharmaceutical patents was the passage by Congress of Public Law 103465, which implements the General Agreement on Trade and Tariffs (GATT) in the United States. Under GATT patent exclusivity, terms will be 20 years from the date of patent application, rather than 17 years from the date of the patent grant. Patents filed on or after June 8, 1995 will be governed by this new rule, which brings the United States in accord with the rest of the world. GATT also has a discretionary transitional feature for patents in effect on June 8, 1995. This permits firms to elect either the 20year term from patent application or the 17year term from patent award. In the case of pharmaceutical patents, however, the Patent and Trademark Office has ruled that one must first add the WaxmanHatch extensions to the original 17year patent period before comparing it with the 20year period from the date of  2 filing.X! 2@ ԍPatent and Trademark Office, "Determination of New Expiration Dates of Certain Patents," Federal Register, Vol. 60, No. 109, June 7, 1995. Since most WaxmanHatch extensions are two years or more, there are very few post1984 NCEs that will receive X01Í ÍX01Í Í  2 transitional benefits under the new GATT rules.Q  ! 2% ԍThe patent office typically takes a few years to grant a patent from the date of initial application. Hence, after also adding WaxmanHatch extensions, most NCEs have a later expiration date under the 17year from grant rule compared to the 20year from'0*(( filing rule.Q X0*((@@ԌWe examined this issue for all 211 NCEs in our sample and found that only 23 NCEs were eligible for GATT discretionary increases under the transitional procedures of the law. Most of these NCEs did not receive WaxmanHatch patent extensions (usually because their average patent life exceeded the 14year upper bound under the 1984 Law). The average GATT extension for the 23 NCEs eligible for such an increase was 11.2 months. Since only 10 percent of the NCEs in our sample are eligible for GATT extensions, and these extensions average less than a year, the impact on the annual average effective patent terms shown in Figures 5 and 6 are negligible (i.e., less than a 0.1 year upward shift on average in the curves shown). NCEs that receive a patent after June 8, 1995, will be covered by a 20year term from the date of application, as instituted by the GATT Agreements. WaxmanHatch extensions then will be added to  2p this expiration date.PpX! 2X ԍP.T.O. Federal Register, June 7, 1995.P Overall, the GATT law could add another year or so of effective patent life to some, as yet unpatented, NCEs compared to what would generally have occurred under the old  2  patent law regime.@ ! 2! ԍFor example, if the U.S. Patents Office took two years to grant a patent to a representative NCE under the prior 17year regime, then a shift to the new regime of a 20year patent from the date of filing means the patent for this NCE would expire one year later in time. The WaxmanHatch time extension would be added to this later expiration date. This would mean a later patent expiration unless the 14year constraint counteracts any such extension. However, a growing number of NCEs are now  0*((@@ being extended to the maximum 14year effective life by WaxmanHatch extensions, and this constraint will work to counteract the GATTinduced increases in patent life in these cases. Furthermore, it will be a decade or so before marketed new drug introductions realize any GATTinduced benefits, given the long gestation times that now exist in pharmaceuticals between patent application and FDA marketing approval.  III. Summary and Conclusionsă As we discussed at the beginning of the paper, the objectives of the WaxmanHatch Act were to facilitate generic entry to spur drug price competition, and also to provide a positive stimulus for drug innovation. In terms of the first objective, the Act has clearly been a tremendous success. The level of generic competition is very different one decade after the Act. In the early 1980s, the level of generic dispensing in the United States hovered around 10 percent. By contrast, in the mid1990s, the level of generic  2  prescribing was approaching 40 percent (Morrison, 1993).I ! 2x ԍFurthermore, there are more than 100 drugs coming off patent between 1992 and the year 2000 with market sales of over $25 billion. As a consequence, one analyst projects that generic drugs will account for more than 65 percent of all new prescriptions by the turn of the century (Shaw, 1992).I The WaxmanHatch Act was not the only explanation for this new world of extensive generic usage. The growth of managed care and other related demand side factors have also played a major role. But the 1984 Act, by making entry into the market relatively easy for`"x0*((@@ generic products, was a necessary condition for the dramatic changes that have occurred in this past decade. Our analysis of major new drugs coming off patent indicates that the degree of generic competition has continued to accelerate in recent years. In particular, the drugs coming off patent since 1991 have experienced unit sales losses to generics of over 50 percent during the first several months of generic competition. This is a much more rapid rate of loss to generics than we observed  2 for similar drugs coming off patent between 1984 and 1989.X ! 2 ԍThe 18 drugs in 1984 to 1989 experienced a 54 percent loss after two years of generic competition, while the 19911993 experienced this rate in less than a year. Consumers and purchasers of pharmaceuticals have clearly been major beneficiaries of the increased generic competition since 1984. While we have not done a precise accounting, the cumulative savings to consumers and payors would be in the billions of dollars. Pharmacists also appear to be significant beneficiaries of the Act, given the larger absolute price markups on generic drugs compared to their brand name counterparts. The effects of the 1984 Act on the incentives for drug innovation are more complex to evaluate. The Act clearly has led to significant patent extensions on recent new drug introductions. For example, the average effective patent life for new drug introduction coming to the market in the 199193 period was 11.7 years, with an average extension of 2.3 years. Moreover, 43 percent of these NCEs had effective lives of 14 years or more (14 years is the upper bound for extensions under the Act).#0*((@@ԌWhile patent periods have increased, there is also a much more rapid rate of sales decline after patent expire. One must balance these two effects in assessing the impacts of the law on new drug returns. Given the many other changes that have also taken place in the market place for new drugs over the past decade, it is not clear what the relevant baseline should be to undertake this type of analysis. One major finding that has emerged from our ongoing studies of the returns to R&D for new drug introductions in the 1970s and 1980s is that product lifetimes are shortening over time. In other words, despite the movement toward longer effective patent lives, product life cycles are plateauing in the prepatent period sooner and they are also subject to much more rapid sales decay rates when  2P patents do expire.XP! 2 ԍFor a summary of our work on the 198084 cohort of new drug introductions and a discussion of preliminary findings for the 198089 cohort, see Grabowski and Vernon (1985). This reflects the increased price sensitivity in the market for pharmaceuticals from managed care, as well as the increased availability of substitute therapies and generic competitors for major pioneer products. In the final analysis, policymakers must consider whether current rules on property rights for pharmaceuticals, devised in a very different market environment, provide the optimal amount of stimulus for the drug innovations of tomorrow. In a prior computer simulation model of the effects of different public policies on innovation (Grabowski and Vernon, 1987), we found that the length of patent protection was a very important policy instrument for#0*((@@ pharmaceuticals, albeit one that was subject to diminishing returns. In particular, we found that the positive stimulus to innovation was disproportionately concentrated in the first decade of market life, with significant diminishing returns as one extended patents to the full nominal life of 17 years. If policymakers are to consider changes in the 1984 Act, we think the first priority should be for products that have little or no effective patent life. In particular, attention should be focused on new drug candidates that must rely on the 5year data exclusivity for their basic protection. We don't know how many potentially useful compounds have not reached the market because of too little patent protection. But we do know that five years is a very short term to recover the hundreds of millions of dollars of R&D investment, required by the typical new drug introduction (DiMasi et al., 1991). The European Community recently enacted patent restoration and data exclusivity policies for pharmaceuticals. One major difference from the United States is that the EC has a data exclusivity period of six to ten years. We believe that U.S. legislators also should consider an increase in the minimum period of protection that new drugs enjoy against generic competition. This should be done in light of the changing economics of drug innovation process over the past decade, including those emanating from the generic competition section of the 1984 Act.#0*((@@ %References X Bernard Bloom, David J. Wierz, and Mark Pauly, "Cost and Price of Comparable Branded and Generic Pharmaceuticals," JAMA, Vol. 256, No. 18, 1986, pp. 25232530.(# X Richard E. Caves, Michael D. Whinston, and Mark A. Hurwitz, "Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry: An Exploratory Analysis," Brookings Papers on Economic Activity: Microeconomics (1991).(# X Joseph DiMasi, Ronald Hansen, Henry Grabowski and Louis Lasagna, "The Cost of Innovation in the Pharmaceutical Industry," Journal of Health Economics, Vol. 10, 1991, pp. 107142.(# X Martin Eisman and William Wardell, "The Decline In Effective Patent Life of New Drugs," Research Management, Vol. 21, January 1981, pp. 1821.(# X Henry Grabowski and John Vernon, "Substitution Laws and Innovation in the Pharmaceutical Industry, Law and Contemporary Problems, Vol. 43, 1979, pp. 4366.(# X Henry Grabowski and John Vernon, The Regulation of Pharmaceuticals: Balancing the Benefits and Risks, Washington: American Enterprise Institute for Public Policy Research, 1983.(# X Henry Grabowski and John Vernon, "Longer Patents for Lower Imitation Barriers: The 1984 Drug Act," American Economic Review, Vol. 76, No. 2, May 1986, pp. 195198.(#$0*((@@ԌX Henry Grabowski and John Vernon, "Pioneers, Imitators and Generics-A Simulation Model of Schumpeterian Competition," Quarterly Journal of Economics, August 1987, pp. 491525.(# X Henry Grabowski and John M. Vernon, "Brand Loyalty, Entry, and Price Competition in Pharmaceuticals after the 1984 Drug Act," Journal of Law and Economics, October 1992.(# X Henry Grabowski and John Vernon, "Prospects for Returns to Pharmaceutical R&D under Health Care Reform," in Strategies in the Pharmaceutical Industry, (Washington, D.C.: AEI Press, 1995).(# X Kenneth I Kaitin and A. Gene Trimble, "Implementation of the Drug Price Competition and Patent Term Restoration of 1984: A Progress Report," Journal of Clinical Research and Drug Development, Vol. 1, 1987, pp. 263275.(# X Jeannine Mandelker, "Get the Most Out of Your PBM," Business and Health, November 1994, pp. 3741.(# X Alison Masson and Robert Steiner, Generic Substitution and Prescription Drug Prices: Economic Effects of State Drug Product Selection Laws, Staff Report of the Bureau of Economics, Federal Trade Commission, Washington, D.C., October 1985.(# X Sylvia Morrison, "Prescription Drug Prices: The Effects of Generics, Formularies and Other Market Changes," Congressional Research Service Report for Congress, August 17, 1983.(##0*((@@ԌX Hemant K. Shaw, "Catalysts for Change: The Impact of Generic Drugs on the Pharmaceutical Industry," Spectrum, Report 38, Decision Resources, Burlington, Mass., December 28, 1992.(# X Sheila Shulman, Brigitta BienzTadmor, Pheak Son Seo, Joseph DiMasi, and Louis Lasagna, "Implementation of the Orphan Act: 19831991," Food and Law Journal, Vol. 47, No. 4, 1992, pp. 363404.(#` 0*((@@    Y ddx !ddx~  Y  "* 'HTable 1 * * Summary of Generic Competition Findings: 19841991 *  ""At Date !of Entry"0One Year /after Entry"?Two Years >-after Entry C p  Average Brand Name Price Index "% $1.0  1.06 (198485) 1.08 (198687) 1.06 (198991) "A1.11 C A1.12 C A1.10p p h Average Generic Price Index"% $1.0 0.77 0.79 0.86H"A0.65 C A0.67 C A0.63p p   Average Ratio of Generic Price to Brand Name Price("#0.63 % #0.59 % #0.61 0.47 0.44 0.49"A0.38 C A0.36 C A0.35p   H Average Generic Market Share in Units"#0.07 % #0.11 % #0.13X 0.32 0.38 0.41X"A0.45 C A0.54 C A0.59   X` hp x (#%'0*,.8135@8:-Generic to @Pioneer >Price Ratio   Wholesale "'A33.38 "4;14.85 "A0.44 C    Retail("'A43.24("4;28.09("A0.65 C    Absolute Markup@"'A 9.86@"4;13.24@"B ԩ C H   Percentage Markup"(30%"589%"B ԩ CH   All prices are measured in cents per pill one year after generic entry took place. See Table A2 for data on the individual drugs included in this sample. \+))@@    Y addx 66 ddx!( ( ( Y H X"* & Table A1 * * Generic Drugs in Sample *XH    Brand Name  Generic Name Date of Generic EntryH  P X Flexeril cyclobenzaprine HCL June 1989P P  Minipress@ prazosin@ June 1989P P  Asendin amoxapine Sept 1989P P @  Proventil/Ventolin albuterol Jan 1990P P   Clinoril0 sulindac0 May 1990P P   Minocin minocycline Sept 1990P P 0 Procardia/Adalat nifedipine Oct 1990P P  Tenormin  atenolol  Aug 1991P P  Imodiump loperamidep Oct 1991P P   Tolectin tolmetin Jan 1992P P p Aventyl/Pamelor nortriptyline Aug 1992P P  Feldene` piroxicam` Aug 1992P P  Cardizem diltiazem HCL Nov 1992P P ` Dolobid diflunisal Nov 1992P P  ViskenP pindololP Nov 1992P P  Orudis ketoprofen Jan 1993P P P Lopid gemfibrozil Feb 1993P P  Sinemet@ caridopa/levodopa@ Feb 1993P P  Corgard nadolol Sept 1993P P @ Naprosyn! naproxen! Oct 1993P P   Xanax0# alprazolam0# Oct 1993P   ! Lopressor$ metoprolol$ Nov 1993  0# Note:` ` The slashes in the brand name list indicate two different brand names of two different cos.      h ddx!( ( ( ddx"DZZZZZ h "   P0#""* & Table A2 * * Drug Store Margins For Pioneer and Generic Products * *$PP   $ "Q Wholesale Prices",Retail Prices"DV $P    P$ "@ Pioneer  Price | T=1 Yr "!Generic "Price !T=1 Yr "*Pioneer +aPrice *%T=1 Yr "4CGeneric 5Price 4T=1 Yr "=Pioneer =Margin "F Generic F!Margin  " M(a) "0 (b) "#(c) ",(d) "63(e) "?(f) "H!(g) I" P   Indomethacin " 23.29 ""13.35 "+a33.13 "524.91 "> 9.84 "GW!11.56P P   Tolazamide8" 30.78""18.298"+a38.488"527.588"> 7.788"GW! 9.29P P   Methyldopa" 15.47"" 9.81"+a19.92"516.24"> 4.45"GW! 6.43P P 8 Chlorpropamide" 28.12"" 5.02"+a36.01"514.23"> 7.89"GW! 9.21P P  Ibuprofen(" 15.51("" 7.81("+a26.74("517.67(">11.23("GW! 9.86P P  Lorazepamx" 27.81x""13.91x"+a40.51x"524.49x">12.70x"GW!10.58P P ( Diazepam" 24.37"" 5.64"+a31.18"516.88"> 6.81"GW!11.24P P x Propranolol" 18.78"" 7.56"+a22.94"514.89"> 4.16"GW! 7.33P P  Metoclopramideh" 21.65h""10.91h"+a30.14h"520.65h"> 8.49h"GW! 9.74P P  Flurazepam" 25.94""14.97"+a38.46"528.45">12.52"GW!13.48P P h Doxepin" 25.66"" 9.90"+a34.21"522.49"> 8.55"GW!12.59P P  HaloperidolX" 59.66X""36.05X"+a73.51X"555.74X">13.85X"GW!19.69P P  Clonindine" 22.19"" 6.34"+a27.65"513.68"> 5.46"GW! 7.34P P X Verapamil" 23.67""12.86"+a31.05"522.13"> 7.38"GW! 9.27P P  CephalexinH "| 137.82H ""50.39H "*%164.67H "4101.33H ">26.85H "GW!50.94P H    Average""   33.38""$3 "14.85""- +a43.24""7 528.09""@A > 9.86""I" GW!13.24H  H  X` hp x (#%'0*,.8135@8: