Skip Navigation

How Increased Competition from Generic Drugs Has Affected Prices and Returns in the Pharmaceutical Industry July 1998

Despite the patent extensions included in the Hatch-Waxman Act, the present discounted value of the average returns from marketing a new drug have fallen by an estimated $27 million, or approximately 12 percent, because of the increase in generic market share since 1984. That calculation, presented in Chapter 4, employs a methodology used by economists Henry Grabowski and John Vernon in various analyses and by the Congressional Budget Office (CBO) in a 1994 study of returns from research and development in the pharmaceutical industry.⁽¹⁾ The calculation is based on estimates obtained from this study's analysis of generic entry after patent expiration. Assumptions similar to the ones in CBO's 1994 study were used to convert the change in the stream of sales revenues to the change in profits.

The key assumptions in the calculation--the rate at which sales revenues eroded before and after the Hatch-Waxman Act and the change in the length of patent protection--are based on analysis of CBO's retail pharmacy data set, data on patent extensions from the Patent and Trademark Office, and a study by the Federal Trade Commission.⁽²⁾ The change in returns is calculated by projecting the value of total U.S. sales revenues in the 12th to 20th year after market introduction for the average drug in CBO's sample of 67 drugs under two scenarios. First, what would sales revenues in those years have been if generic market share (for nonantibiotic drugs) were at its pre-1984 average? And second, what would sales revenues in those years be with a 2.8-year patent extension and increased generic market share at the end of year 14, as is the case today?

From those two revenue streams, the change in profits in years 12 to 20 is calculated assuming a marginal cost of production equal to 25 percent of the brand-name wholesale price. That assumption is well grounded in the literature on the pharmaceutical industry.⁽³⁾ Since the appropriate measure of returns is after-tax profits, a marginal tax rate of 35 percent is also applied. Thus, an increase in sales revenues of $1 would add 49 cents to after-tax profits in a given year.⁽⁴⁾ The change in profits is then discounted to the date of market introduction using a real interest rate of 10 percent (consistent with previous studies that have measured the average returns from marketing new drugs).
 

Formulas

Because of the patent-term extensions available after 1984 and the delay between patent expiration and generic entry that existed before 1984, the sales streams in the two scenarios do not begin to diverge until year 14. The formula used for converting the difference between the pre- and post-1984 sales streams into profits (discounted to the date of market introduction) is:
 

To obtain the pre- and post-1984 streams of sales revenues, an assumption is needed about the rate at which those revenues would erode without generic entry. For both streams, sales revenues were assumed to decline gradually starting in year 14 because of competition from other, improved innovator drugs. That erosion rate was assumed to be 6 percent in year 14 and to increase by 2 percentage points each year thereafter. The formula for sales revenue erosion caused by competition from other innovator drugs starting in year 14 is:
 

That revenue stream is then further reduced depending on the size of the generic market. The bigger the generic market share, the smaller will be the sales revenues for the average innovator drug. The formula used to project the revenue stream, accounting for generic entry, is:
 

Generic Market Share Before 1984

Before the Hatch-Waxman Act, generic market share was very small for most multiple-source drugs, with the exception of antibiotics. Generic market share averaged just 12.7 percent for 29 multiple-source drugs that were among the top 100, rated by total U.S. sales revenues, in 1980.⁽⁵⁾ Those were drugs for which generic entry had occurred, however. Actual generic market share for the average brand-name drug before 1984 was smaller than that after accounting for cases in which generic entry did not occur.

Besides generic market share being small, the probability of generic entry was low for an off-patent brand-name drug before 1984. After excluding antibiotics and drugs approved before 1962, only 35 percent of the remaining top 200 drugs had generic versions available in 1983.⁽⁶⁾ A few of those drugs had had their patent expire in 1980 or later; hence, the overall probability of generic entry at the average time it occurred (three years after patent expiration) was assumed to be slightly higher, 40 percent (see Table C-1). As a result, average generic market share for all multiple-source drugs was assumed to be 5.1 percent (40 percent of 12.7 percent), although that figure would be a bit smaller in the first year after generic entry.
 

Generic Market Share After 1984

CBO assumed that in the post-Hatch-Waxman period, generic entry normally occurs within 1.2 months of patent expiration. That figure resulted from examining 17 top-selling nonantibiotic drugs whose patents expired between 1990 and 1993. For 14 of those drugs, the average delay between patent expiration and generic entry was just over one month. (The date of generic entry for those drugs was included in a paper by Grabowski and Vernon; CBO obtained the date of patent expiration from the Food and Drug Administration's so-called Orange Book for 1990).⁽⁷⁾ For the other three drugs, generic entry took 17 to 21 months after patent expiration; but according to an official of the Food and Drug Administration (FDA), that delay occurred largely because the agency was unable to evaluate those applications quickly since it was recovering from a scandal in the generic drug industry.⁽⁸⁾

CBO's assumption about the size of the generic market shortly after patent expiration and generic entry is based on an analysis of 21 innovator drugs in the retail pharmacy data set that first faced generic competition between 1991 and 1993. Generic sales constituted an average of 44.2 percent of total prescription sales for those drugs during the first full calendar year after generic entry.⁽⁹⁾ Since that figure is based only on cases in which generic entry occurred, CBO adjusted it by the estimated probability of such entry--calculated to be 91.5 percent (see Box C-1). As a result, the average generic market share in the year following patent expiration, accounting for cases in which generic entry does not occur, is estimated to be about 40 percent.
 

By the time three years have elapsed since generic entry, the average generic market share for a drug is assumed to have reached 60 percent. CBO estimated that figure as follows. Overall generic market share--calculated as the volume of generic countable units sold to all purchasers in the United States divided by the volume of all drugs sold, including single-source drugs--was 40.4 percent in 1994, according to IMS America. (Note that this figure for 1994 generic market share is lower than the 50.5 percent figure in Box C-1 because it is taken as a percentage of all drug sales rather than just sales of multiple-source drugs.) Based on the retail pharmacy data set, CBO estimated that including all dosage forms in that average, rather than just those that are easily countable, such as tablets and capsules, reduces generic market share to 38.2 percent.⁽¹⁰⁾ To calculate average generic market share for drugs that have been off patent for three or more years, the 38.2 percent figure was divided by 66.7 percent, the share that off-patent drugs constituted of the retail pharmacy data set in 1994. As a result, CBO estimated that generic sales represented 57.3 percent of all sales of multiple-source and off-patent single-source drugs in 1994.⁽¹¹⁾ Since generic market share has continued to increase slightly since 1994, and since older drugs would have a slightly higher generic share than the market average (which includes drugs that recently went off patent), CBO adjusted that estimate of average generic market share upward to 60 percent.⁽¹²⁾

The figure for generic market share in the second year after patent expiration, 50 percent, is simply an average of the figures for the first and third years.

In CBO's calculations of generic market share, the estimated probability of generic entry (91.5 percent) helps to account for the cases in which generic entry is delayed by a year or more. In the first year following patent expiration, generic market share equals 44.2 percent multiplied by 91.5 percent, or 40 percent. In the third year after patent expiration and later, the cases in which generic entry did not occur were incorporated into the calculation of a generic market share of 60 percent. How sensitive CBO's calculation of the change in returns from marketing is to those estimated generic market shares is analyzed below.

The formulas used to project generic market share based on this analysis are shown in Table C-2. The first year of patent expiration is split between years 14 and 15 of a drug's product life. Since generic entry is assumed to occur at the very end of year 14, in that year generic market share is equal to only 10 percent of 44.2 percent multiplied by 91.5 percent, which is 4 percent. In year 15, generic market share is a weighted average of generic market share in the first year after patent expiration (90 percent) and the second year after patent expiration (10 percent). The average generic market share for year 15 is therefore 41 percent.
 

Sensitivity Analysis

CBO examined the sensitivity of its estimate of the decline in the present discounted value of the average returns from marketing a new drug ($27 million) to the assumptions used to construct the pre- and post-1984 streams of sales revenues. The results indicate that the estimate is little affected by modest changes in the key assumptions (see Table C-3).
 

If, in constructing the pre-1984 sales stream, CBO assumed that generic drugs took four years instead of three to enter the market after patent expiration, the estimated decline in returns would be $28 million, just $1 million different. That change is small because the size of the pre-1984 generic market was small, so postponing generic entry by another year in that period does not have much effect on the basic result.

The effect would be greater if generic entry was further postponed in the post-1984 period (since generic market share is higher then), but the data that underlie CBO's estimate of a 2.8-year average postponement under the Hatch-Waxman Act are solid. If the average length of a patent extension was six months shorter, returns would fall by an additional $5 million. If the average length was six months longer, the decline in returns would be $4 million less. However, the data on patent-term extensions obtained for all drugs approved between 1992 and 1995 make it unlikely that the estimated average length of an extension would be off by as much as six months.

CBO's basic result is not very sensitive to a small increase in the size of the generic market. For example, if the post-1984 generic market share was 45 percent in the first year after generic entry, rising to 65 percent in the third year and beyond, the decline in returns would be $30 million--only $3 million more than the base case. Those alternative assumptions are based on what might be a reasonable upper bound for current levels of generic market share.

CBO assumed that the marginal cost of producing another unit of a prescription drug was 25 percent of its brand-name price. Varying the marginal cost from 20 percent to 30 percent of the brand-name price causes the total decline in returns to vary between $25 million and $29 million. Thus, CBO's estimate is not particularly sensitive to reasonable variations in incremental unit costs.

As a drug becomes obsolete and its efficacy is surpassed by that of newer innovator drugs, its sales revenues gradually erode. That erosion rate was assumed to be 6 percent in year 14 and 8 percent in year 15, increasing by 2 percentage points each year thereafter. If CBO had used a slightly slower rate of revenue erosion caused by product obsolescence--starting at 5 percent in year 14 and increasing by 1 percentage point each subsequent year--the total decline in returns would be an estimated $30 million. By contrast, with a faster erosion rate--6 percent in year 14, increasing by 3 percentage points each year thereafter--returns would decline by $25 million. Hence, the estimate is fairly robust to that assumption as well.

As discussed in Chapter 3, the extent to which brand-name prices respond to generic entry is unclear from previous studies. CBO's base case assumes that those prices do not respond to generic entry. If brand-name prices did change because of generic competition, they would primarily affect the profit stream in the post-1984 scenario, since generic market share was so small before 1984. If, because of increased discounting, the average brand-name price was 5 percent lower in each year after generic entry in the post-1984 scenario, the returns from marketing a new drug would fall by $29 million, a difference of $2 million. If, conversely, the average brand-name price was 5 percent higher in that period after generic entry, estimated returns would fall by $25 million. Thus, CBO's calculation is not highly sensitive to any effect that generic entry might have on brand-name prices.

An important number on which the calculation depends is sales revenues in year 13 (the average drug's peak year, before product obsolescence and generic entry occur). According to CBO's data set, those revenues averaged $139.2 million in 1990 dollars.⁽¹³⁾ CBO's estimate of the change in returns is not very sensitive to modest changes in those revenues. For example, if sales revenues in year 13 were 10 percent lower, the estimated decline in returns would be $24 million. If sales revenues in year 13 were 10 percent higher, the estimated decline in returns would be $30 million. Of course, if revenues were 10 percent lower or higher in all years leading up to year 13, then total returns would also be lower or higher than the assumed $210 million to $230 million. But even accounting for the corresponding change in total returns, the result (taken as a percentage of the total expected returns from marketing a new drug) would remain a decline of about 12 percent, on average.

1. Henry Grabowski and John Vernon, "Returns to R&D on New Drug Introductions in the 1980s," Journal of Health Economics (1994); Grabowksi and Vernon, "Longer Patents for Lower Imitation Barriers: The 1984 Drug Act," American Economic Review (May 1986); and Congressional Budget Office, How Health Care Reform Affects Pharmaceutical Research and Development (June 1994).

2. Alison Masson and Robert Steiner, Generic Substitution and Prescription Drug Prices: Economic Effects of State Drug Product Selection Laws (Federal Trade Commission, 1985).

3. See, for example, Office of Technology Assessment, Pharmaceutical R&D: Costs, Risks and Rewards (February 1993), p. 79. Also see CBO, How Health Care Reform Affects Pharmaceutical Research and Development, pp. 51-53.

4. Because (1 - 0.25)(1 - 0.35) = 0.49.

5. The 12.7 percent average was calculated based on Table A5-1 in Masson and Steiner, Generic Substitution and Prescription Drug Prices. The sample in that report contained 45 multiple-source drugs. Ten were antibiotics, and six others were eliminated because they were still under patent, had minimal generic sales, or were only available under a generic name.

6. Grabowski and Vernon, "Longer Patents for Lower Imitation Barriers," pp. 195-198.

7. See Henry Grabowski and John Vernon, "Longer Patents for Increased Generic Competition in the U.S." PharmacoEconomics (1996), Table 1, p. 112; and Department of Health and Human Services, Food and Drug Administration, Approved Drug Products with Therapeutic Equivalence Evaluations (1990).

8. Personal communication with an FDA official on March 26, 1998.

9. The unweighted average generic market share for the 21 drugs was 43 percent. Weighting that average (a volume measure) by the value of the drugs' retail pharmacy sales revenues in 1991 (thus giving higher-selling drugs a greater emphasis) yields an average generic market share of 44.2 percent.

10. Limiting the calculation only to tablets and capsules raises the average generic market share calculated from the retail pharmacy data set by 2.2 percentage points.

11. That figure has already been adjusted to account for cases in which generic entry was prevented, since the sales of single-source, off-patent brand-name drugs were accounted for in the 66.7 percent.

12. IMS America estimated that overall generic market share in 1996 was 42.6 percent. Adjusting that figure from mainly tablets and capsules to all dosage forms would imply a market share of 40.4 percent. Then, assuming the same split between brand-name and generic drugs in 1996 as in 1994 would yield a generic market share of 60.5 percent for drugs off patent.

13. Average U.S. sales for the 67 drugs in CBO's sample were $139.2 million in year 11 and were assumed to continue at that level through year 13.

---