Skip Navigation

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

The federal government has competing policy objectives with respect to the pricing of prescription drugs. On the one hand, it wants to ensure that companies have enough incentive to invest in researching and developing innovative drugs. On the other hand, it wants to discourage them from charging excessively high prices. In general, the government achieves the first goal through a patent system that grants market exclusivity for a limited period of time, allowing companies to recoup their investment in R&D. For the second goal, it relies on competition between similar drugs to hold prices down.

This chapter examines price competition among manufacturers in the pharmaceutical market, including the impact of the dramatic growth in the generic drug industry since 1984. Such competition comes in three main forms: between brand-name drugs in the same therapeutic class, between brand-name drugs and their generic counterparts, and between different generic versions of the same drug. The pharmaceutical industry is also affected by other types of competition, such as the substitution that sometimes occurs between prescription drugs and other forms of medical treatment. However, the conditions under which prescription drugs can be substituted for other medical procedures are outside the scope of this study.

The patent system provides a period of protection during which manufacturers of innovator drugs can charge relatively high prices, earning profits that enable them to compensate for the costs of a drug's discovery and development. Although patents prevent other manufacturers from producing the same drug, they do not prevent manufacturers with a similar but slightly different drug from also obtaining a patent and entering the market. Limited empirical evidence suggests that the availability of several similar brand-name drugs tends to slow the rate of price growth, even before generic copies become available.

The dramatic rise in generic sales since 1984 has held down average prices for drugs that are no longer protected by a patent. However, those lower prices tend not to result from reductions in the price of the original brand-name drug when it begins facing competition from generic drugs. Rather, average prices fall primarily because consumers switch from the higher-priced innovator drug to the lower-priced generics. To be on the receiving end of that switch, generic manufacturers compete with each other intensely in the area of price, partly because they sell identical products.

The increased use of generic drugs has kept total spending on prescription drugs below what it might otherwise have been. Considering only drugs sold through retail pharmacies, the Congressional Budget Office (CBO) estimates that the purchase of generic drugs reduced the cost of prescriptions (at retail prices) by roughly $8 billion to $10 billion in 1994. That estimate assumes that all generic prescriptions dispensed would have been filled with a higher-priced brand-name drug if the generic was not available.

Much of the analysis in this chapter relies on a set of data that represents 70 percent of prescription drug sales at retail pharmacies in the United States in 1993 and 1994. Roughly half of all prescription drugs are channeled through retail pharmacies (see Figure 2). Thus, the data set represents about 35 percent of all drug sales in the United States in those years. The data include total dollars spent on each dosage form (tablet, capsule, liquid, and so forth) of 454 brand-name drugs, as well as total spending on generic versions of the brand-name drugs whose patents have expired. (For more information about the data, see Appendix A.)
 

Figure 2. Channels of Distribution for Prescription Drugs

SOURCE: Congressional Budget Office based on Micky Smith, Pharmaceutical Marketing Strategy and Cases (New York: Pharmaceutical Products Press, 1991), Chapter 3; Boston Consulting Group, The Changing Environment for U.S. Pharmaceuticals (Boston: Boston Consulting Group, April 1993); and Pharmaceutical Research and Manufacturers of America, 1997 Industry Profile (Washington, D.C.: PhRMA, March 1997), p. 31.

NOTES: Figures in parentheses represent shares of the prescription drug market in 1996, calculated as a percentage of total U.S. sales at manufacturer prices.

HMOs = health maintenance organizations.

a. Some chain-store pharmacies buy directly from the manufacturer.

b. Some mail-order pharmacies go through a wholesaler.

The unit used to measure quantity in the retail pharmacy data is the prescription. That unit can lead to measurement errors, however, since different prescriptions for the same drug come in different sizes. (For example, one prescription may be filled with 30 pills and another with 100 pills.) A statistical bias would occur if more pills were dispensed per prescription, on average, for a generic drug than for its brand-name counterpart. That bias would lead to underestimating the price difference between brand-name and generic drugs. But the bias could run in either direction. Without a better measure of quantity, part of the analysis in this chapter relies on the number of prescriptions to estimate sales volume and to calculate average unit prices. Implicitly, those estimates assume that, in general, prescriptions for a brand-name drug and for its generic equivalent have roughly the same average number of dosage units (such as tablets). All estimates that rely on average prescription prices are based only on tablet and capsule formulations, which constitute 87 percent of sales in the retail pharmacy data set. Those dosage forms yield more reliable average prescription prices.
 

Competition Among Brand-Name Drugs

In 1994, 83 percent of retail pharmacies' total revenues from selling prescription drugs came from innovator drugs (see Table 1). Those brand-name drugs also accounted for 64 percent of all prescriptions dispensed. Single-source innovator drugs--which, by definition, do not yet face generic competition--made up half of retail pharmacies' revenues from the sale of prescription drugs. Because innovator drugs constitute such a large share of pharmacy sales, the extent to which their manufacturers compete on the basis of price has important implications for consumers.
 

In general, the higher prices charged for brand-name drugs allow firms to recoup their investment in a drug's discovery and development. Studies have found that, on average, discovering and developing a drug takes 11 to 12 years and costs about $200 million per successful product (in 1990 dollars).⁽¹⁾ That $200 million figure includes the cost of drugs that never make it to market; it also accounts for the cost of capital--that is, the cost of waiting for a return until the drug is introduced. Actual drug development costs may be higher today if, for example, the cost of conducting clinical trials has increased. Conversely, costs may be lower if the failure rate of drugs that go into clinical trials has declined.⁽²⁾

The stream of after-tax profits over the life of a typical innovator drug follows an up-and-down pattern (see Figure 3). The first 11 to 12 years show a negative cash flow while the drug is being developed, undergoing testing, and awaiting approval. Over the next 20 years, as the drug is marketed, it earns back a return on the investment in its research and development. According to two studies, that profit stream has an average present value of $220 million to $230 million (in 1990 dollars, after deducting manufacturing, advertising, distribution, and other non-R&D-related costs, discounted to the date of market introduction)--which more than compensates for the $200 million in average capitalized costs of drug development.⁽³⁾ Those studies estimate that for innovator drugs introduced in the early 1980s, after-tax profits exceeded development costs by $22 million to $36 million, on average (in 1990 dollars, where returns are discounted and costs are capitalized to the date of market introduction). Since the returns from selling new drugs are highly skewed--a few drugs earn very large profits, whereas others may only cover the cost of their own development--that average encompasses both a few big winners and some marginally profitable drugs.
 

Figure 3. Change in the Profit Stream for a Typical Innovator Drug

SOURCE: Congressional Budget Office based in part on Henry G. Grabowski and John M. Vernon, "Returns to R&D on New Drug Introductions in the 1980s," Journal of Health Economics, vol. 13, no. 4 (December 1994).

NOTE: R&D = research and development.

The FDA Approval Process

Much of the capitalized cost of drug development can be attributed to the length of the discovery, development, and approval process. That process includes five distinct phases, the first of which is screening and discovery. Recent advances in biomedical research appear to have increased productivity in the discovery phase by yielding new "targets" (such as enzymes) against which a chemical can be tested for interactions. A process called high-throughput screening allows hundreds of chemicals to be tested quickly against a single target. After finding a drug candidate that interacts with the target, the manufacturer checks the drug for toxicity and tests it in animals. If the drug still looks promising, the company files an investigational new drug application with the Food and Drug Administration in order to begin testing the compound in humans. (Testing can begin 30 days after the application is filed.) Between 1980 and 1992, that screening and discovery phase (including preclinical testing) took an average of two to four years.⁽⁴⁾

The clinical trials that follow are divided into three phases. Phase I tests the new compound on fewer than 100 volunteers (usually healthy people) to determine safe dosage levels and toxicity. Phase II tests the drug on 50 to 200 people who have the disease the drug is designed to combat in order to determine both safety and efficacy. Phase III tests the drug on thousands of people to see whether the benefits are statistically significant.⁽⁵⁾ The FDA usually requires two controlled clinical trials in humans (Phase III studies) before approving a new drug.⁽⁶⁾ Those trials establish effectiveness, optimal dosage forms, and possible side effects. They can also detect adverse reactions at that stage. Companies often consult with the FDA when designing their clinical tests. After a company believes it has gathered sufficient evidence in Phase III testing, it files a new drug application with the Food and Drug Administration.

Making the drug-approval process as quick and efficient as possible without sacrificing standards of safety and efficacy benefits both the public and pharmaceutical manufacturers. Those were the goals of the 1992 Prescription Drug User Fee Act (PDUFA). Meeting those goals is not a simple task, however, since "inevitably there is a trade-off between speed and certainty" about a drug's safety and effectiveness.⁽⁷⁾ The PDUFA imposed fees on pharmaceutical manufacturers when they submit a new drug application for FDA approval. In 1997, those fees totaled $205,000 for a full NDA requiring clinical data for approval. Other types of fees paid by firms that filed NDAs include an annual fee on their manufacturing establishments and an annual fee for the drugs they currently have on the market.⁽⁸⁾

The FDA has used those fees to hire more reviewers and accelerate the approval process. The agency reports that it has eliminated the backlog of applications that were awaiting approval and has sped up approval for applications filed since 1992. According to the FDA, drug applications approved between 1991 and 1992 (that included a chemical never before approved) had a median review time of about 22 months. For applications approved in 1994 and 1995, the median review time was down to about 15 months, falling to 12 months in 1996.⁽⁹⁾

By law, however, the FDA is required to approve all new drug applications within 180 days (or a longer period if agreed on with the applicant).⁽¹⁰⁾ In complying with the PDUFA, the FDA has set a target date of one year for all such applications. It reports that at least 95 percent of the 106 new drug applications filed in fiscal year 1995 met that goal.⁽¹¹⁾

The total time a drug spends in development, however, does not appear to have changed much. Steering a new drug through clinical testing in humans to final FDA approval took eight to nine years for drugs approved between 1980 and 1992, according to one study.⁽¹²⁾ CBO found similar results for 186 drugs approved between 1984 and 1995 that obtained patent extensions under the Hatch-Waxman Act. Those drugs spent an average of 5.4 years in the clinical testing phase (see Table 2). The NDA approval phase took another 2.9 years, on average, bringing the total development time after clinical testing began to 8.2 years. For drugs approved in 1994 and 1995, the NDA approval phase was shorter than that average but the clinical testing phase was longer. That suggests that faster NDA review times in recent years may have been partially offset by longer clinical testing periods. However, more data are required to assess whether that is indeed the case.
 

Last year, the Congress passed the Food and Drug Administration Modernization Act of 1997, which made a variety of changes affecting how the FDA regulates food, medical devices, and prescription drugs. Some of those changes could speed up the approval process for innovator drugs. Under the act, the FDA must formulate a plan to reach compliance with the 180-day limit on NDA approvals and other existing time limits.⁽¹³⁾ That plan could further reduce the average approval time for a new drug application if the FDA received enough funding to carry it out. (For example, the agency would probably need to hire more staff.)

The 1997 act also attempts to decrease the time needed for conducting clinical tests by encouraging cooperation between the FDA and pharmaceutical companies. For example, once the FDA has approved an investigational new drug application, its officials are required to meet with the applicant (on written request) to agree on the size and design of the clinical studies necessary for final FDA approval.⁽¹⁴⁾

Faster approval of new drugs increases the returns that those drugs earn. For example, speeding up the FDA approval phase by one year would boost the average profits from marketing a new drug by about $22 million (at a present discounted value in 1990 dollars).⁽¹⁵⁾ That estimate assumes that the approval is accelerated entirely because the FDA reviews applications and test results more quickly, so the timing of outlays in the R&D process does not change. The estimated benefits arise because firms begin earning a profit on their new drug one year earlier. Such a change would nearly double the estimated $22 million to $36 million by which after-tax profits from selling a brand-name drug exceed drug development costs, on average.⁽¹⁶⁾ As a point of comparison, extending the patent on a prescription drug by one year would increase the present discounted value of its returns by substantially less--about $12 million, on average. An additional effect of faster approvals would be increased competition in the pharmaceutical market as new brand-name drugs were introduced more quickly, providing more competition for existing ones.

"Me-Too" Drugs

Although patents prevent other companies from producing exactly the same drug claimed in the patent, they usually do not prevent the introduction of similar but slightly differentiated drugs. In many cases, several different chemical entities can be found that use the same basic mechanism to treat an illness. Since patents are frequently obtained on a specific chemical formulation, not on a therapeutic mechanism, many patented products are "functionally similar."⁽¹⁷⁾ Thus, a breakthrough drug--the first innovator drug to use a particular therapeutic mechanism--may have only one to six years, at most, of pure market exclusivity before a similar patented drug (sometimes called a "me-too" drug) is approved by the FDA. Of 13 therapeutic categories that CBO examined for this study, the first me-too drug entered the market within one year in six cases and within two to six years in another six cases.⁽¹⁸⁾

Consider the example of Tagamet, a breakthrough drug in antiulcer therapies that was introduced in 1977. Tagamet was the first drug to relieve ulcers by blocking the histamine 2 (H2) receptors in the lining of the stomach from stimulating acid production by the parietal cells. Such treatment is generally superior to antacids, which only neutralize stomach acid, as well as to anticholinergic drugs, which block acid production but often have more severe side effects.⁽¹⁹⁾ Six years after Tagamet became available, a second H2 antagonist, Zantac, was approved; it eventually became the largest-selling drug in both the United States and the world. By 1989, two additional H2 antagonists, Pepcid and Axid, were available. Thus, four slightly different drugs using the same therapeutic mechanism (blocking the H2 receptor) were all patentable, and the breakthrough drug had only six years of market exclusivity before being challenged by a competitor using a similar compound.

The Economics Behind the Pricing of Innovator Drugs

Although me-too drugs do not offer a novel treatment, they may have fewer side effects and may treat some patients more effectively than the original breakthrough drug. In addition, me-too drugs create more competition in the market by ending the breakthrough drug's monopoly on its method of treatment. That added competition generally keeps the manufacturer of the breakthrough drug from raising its price as quickly as would otherwise be the case.

According to economic theory, both demand and production costs play a role in determining the price of a drug. The line that illustrates demand for a manufacturer's output (known as a demand curve) slopes downward because people will buy more as the price declines (see Figure 4). For example, if the manufacturer's price decreases from p₁ to p₂, then the quantity that the company can sell increases from q₁ to q₂. It is profitable for the manufacturer to lower the price from p₁ to p₂ only if the increase in profits from the larger quantity sold (represented by shaded area B) more than compensates for the loss in profits from the lower price charged on the first q₁ units sold (represented by shaded area A). In this example, the manufacturer would continue to lower the price until it could no longer profit from doing so.⁽²⁰⁾ The profit-maximizing, or equilibrium, price will exceed the cost of producing another unit of the drug, and the profits earned from selling at that price (represented by areas B plus C, if p₂ is the equilibrium price) provide the incentive for companies to invest in drug development.
 

Figure 4. Choosing a Profit-Maximizing Price for a Drug

SOURCE: Congressional Budget Office.

NOTE: According to this hypothetical demand curve, when the price of a drug declines from p1 to p2, the quantity sold increases from q1 to q2. Area A represents the loss in profits when the price falls from p1 to p2, and area B represents the increase in profits because a greater quantity is sold. Drug companies can increase their profits by lowering price so long as area B is larger than area A. At price p2, total profits equal area B plus area C.

When a breakthrough drug is introduced, by definition it has no close substitutes on the market. Demand for the drug is therefore fairly insensitive to price, since no alternative treatment of equal quality and effectiveness exists. (In other words, the drug has a much steeper demand curve, and a given percentage change in its price is associated with a smaller percentage change in the quantity sold.)

Over time, advertising, "detailing" (visits by representatives of the manufacturer to health care professionals), and articles in medical journals disseminate information to doctors about the new treatment. As the breakthrough drug becomes more widely known, demand for it increases. (Graphically speaking, the demand curve shifts to the right, meaning that at any given price, the manufacturer can sell more of the drug.) At that point, the quantity of the drug sold increases, and its equilibrium price usually rises.

Later, when me-too drugs enter the market, demand for the breakthrough drug becomes more sensitive to price, since close substitutes are now available. At that point, an increase in the price of the breakthrough drug will prompt some purchasers to switch to the substitutes. Advertising and detailing of the new me-too drugs may also cause some customers to switch. Publicity for me-too drugs can also boost demand for the treatment in general. But although the overall market for the treatment may grow, such growth may not offset the sales that the breakthrough drug loses to its new competitors. As the market becomes split among several drugs, demand for the breakthrough drug could shrink and become more sensitive to price. As a result, the price of the breakthrough drug could theoretically decline.

Empirical Evidence About the Pricing of Innovator Drugs

Studies of competition among similar brand-name drugs show that manufacturers compete through prices as well as through advertising and product quality. Most of the empirical studies that look at prices of brand-name drugs are based either on list prices or on average prices paid on invoices to pharmacies and hospitals. Neither of those prices represents an actual transaction price, however. No purchaser pays the list price, although it serves as an important signal since it is a published price observed by all buyers.⁽²¹⁾ The average invoice price is much closer to an actual transaction price, but it does not include rebates or discounts that do not appear on the invoice. Since neither price captures the full impact of discounting, studies that rely on those prices underestimate to some extent the level of price competition among brand-name drugs. Those are the only prices widely available to researchers, however, so they are the ones generally used for analyses.

CBO examined the list prices of breakthrough and me-too drugs over time for five therapeutic classes.⁽²²⁾ In four of the five, the list price of the breakthrough product continued to increase in real terms--that is, by more than just the effects of inflation--after the entry of one or more me-too products.⁽²³⁾ In only one case (that of fluoroquinolone anti-infectives) did the breakthrough drug lower its list price in real terms after the first me-too drug entered the market.

A study by John Lu and William Comanor also found that the average list price of brand-name drugs continues to rise faster than inflation after the introduction of a me-too competitor.⁽²⁴⁾ For 13 drugs that received an A rating from the FDA (as most innovative), the average inflation-adjusted list price after eight years on the market was 7 percent above the launch price. For 48 B-rated drugs (slightly less innovative), the inflation-adjusted list price was 32 percent higher, on average, eight years after launch.

That same study also found that although prices continued to increase, the rate of increase was slower for those drugs that had more brand-name competitors on the market. The introductory price also tended to be lower when more similar brand-name drugs were already on the market. Those findings suggest that the rate of price increase is slowed by competition between brand-name drugs.

A breakthrough drug has an advantage over its me-too competitors in that doctors become experienced with it first and are usually hesitant to try a new drug unless it is seen to be more effective or have fewer side effects. New me-too drugs that offer small advantages over competitors may be sold at a lower price initially; then, as they become more widely accepted, their price rises more quickly.⁽²⁵⁾ That may partially explain why the list prices of C-rated drugs (least innovative) tend to increase much more rapidly over time than the list prices of their more innovative competitors. Lu and Comanor found that for a sample of 69 C-rated drugs, the average inflation-adjusted list price after eight years on the market was 62 percent above the launch price. That high price increase occurred although those drugs were launched at roughly the same price as their closest competitors, on average.

Price competition among similar innovator drugs is softened because products are differentiated. It is also softened because entry in the pharmaceutical industry is limited by patent protection and the FDA approval process. Still, companies have an incentive to continue to enter the market with similar brand-name drugs until profits are driven down to a normal (competitive) rate of return that adequately compensates for the risk of investing in drug development. One economist has asserted, based on discussions with industry executives, that more me-too drugs are not developed because they would not be profitable given the high development costs.⁽²⁶⁾ Companies will choose to develop a brand-name drug similar to others on the market only if they believe that the market is not already saturated, or that their drug may have some quality advantage (such as fewer side effects or greater efficacy) that could enable it to compete effectively and earn profits that more than cover the development costs. Competition should result in firms' earning close to a normal rate of return to their R&D investment, on average.

Using average invoice prices, economist Scott Stern found that cross-price elasticities (a measure of buyers' sensitivity to price differences between similar brand-name drugs) in four therapeutic classes were consistent with the assertion that brand-name drugs compete in price.⁽²⁷⁾ His estimates of price sensitivity were not consistent with the assertion that firms collude to maintain prices as high as what would be charged if a single company produced all of the products. Several other studies have also found that the price differences between patented pharmaceutical products can largely be accounted for by differences in quality, such as side effects and therapeutic effectiveness.⁽²⁸⁾

Barriers to Entry and Market Concentration

Competition between brand-name drugs may be limited not only by patent protection but also by the advantages that large drug companies have in marketing and in the FDA approval process. One of the key ways in which firms compete for market share (other than through price) is by advertising. Promotional spending for a brand-name drug can run as high as 20 percent of total sales. In 1989, three-quarters of promotional outlays went toward detailing--financing a large sales force that promoted the firm's entire product line directly to health care professionals.⁽²⁹⁾ The ability to spread those promotional costs across a large product line is beneficial in that type of marketing, giving big firms an advantage.⁽³⁰⁾ They also appear to enjoy an advantage in the drug-approval process: the General Accounting Office found that NDAs from the most experienced sponsors were three times more likely to be approved than those from the least experienced sponsors.⁽³¹⁾

Perhaps because of the advantages enjoyed by large firms, many new drugs are marketed by a company that did not discover them.⁽³²⁾ Of all chemical entities that began clinical testing between 1979 and 1986, around 29 percent were acquired by another company rather than self-originated (see Table 3). And of the new chemical entities that were approved for marketing during those years, 40 percent were acquired rather than self-originated.
 

At first glance, the pharmaceutical industry does not appear to be highly concentrated. The four largest manufacturers of innovative drugs each accounted for only 6 percent to 7 percent of total U.S. pharmaceutical sales in 1994. And the top 10 companies together shared just 56 percent of the market.⁽³³⁾

When pharmaceutical sales are divided into narrower submarkets, in which products are grouped only with their immediate competitors, much higher concentration becomes apparent. CBO's retail pharmacy data set divides drugs into narrowly defined therapeutic classes. (For more information about how those classes are defined, see Box 3.) The data cover 66 therapeutic classes that together represent about 70 percent of the total retail pharmacy sales revenues in the United States from 1991 to 1994. In just over half of those classes, the top three innovator drugs accounted for 80 percent or more of retail pharmacy sales in their class (see Figure 5).⁽³⁴⁾ In only nine of the classes did the top three innovator drugs make up less than 50 percent of their pharmacy market.
 

Figure 5. Market Share of the Top Three Innovator Drugs in 66 Therapeutic Classes, 1994

SOURCE: Congressional Budget Office based on tabulations of retail pharmacy sales data from Scott-Levin.

NOTE: Market share is calculated as the total sales (valued at retail prices) of the top three innovator drugs in a therapeutic class divided by the total sales of all drugs (both brand-name and generic) in that class.

The level of market concentration varies widely among therapeutic classes, however, with concentration reduced by the availability of several different brand-name drugs and by generic entry. Generally, the most concentrated classes in the retail pharmacy data set had four or fewer innovator drugs, none of which were available in generic form. In the 18 least concentrated therapeutic classes, at least one of the three top-selling innovator drugs had a generic version available. And 14 of those 18 least concentrated classes had nine or more innovator drugs.
 

Factors That Determine Discounts on Brand-Name Drugs

Different purchasers pay different prices for brand-name prescription drugs. Such discounting, which economists refer to as price discrimination, may be an important mechanism for aiding price competition in the pharmaceutical market.⁽³⁵⁾ It rewards institutional purchasers that organize their patient base through formularies so as to encourage the use of less costly drugs, when possible. Prohibiting or limiting discounts, as some people have called for, could decrease price competition.

A statistical analysis of pharmaceutical prices shows that purchasers tend to obtain higher discounts from manufacturers on brand-name drugs when generic substitutes are available and when a greater number of therapeutically similar brand-name drugs are available. That finding suggests that manufacturers' discounts are a response to competitive market conditions. When a variety of similar drugs are available, the purchaser has more opportunities to switch, which can be used as leverage in negotiating discounts.

The Economic Theory Behind Discounting

If companies practice price discrimination, those purchasers least sensitive to price pay the most. In today's market for outpatient prescription drugs, that means people who have no insurance coverage for drugs, or third-party payers that do not use a formulary to manage their outpatient drug benefits, pay the highest prices for brand-name drugs. Differences in price result because manufacturers apply typical profit-maximizing strategies based on the price sensitivity of buyers. According to economic theory, no purchaser pays a higher price to make up for the discounts offered to somebody else. Instead, each pays the price dictated by his or her price sensitivity.⁽³⁶⁾

Manufacturers offer discounts on brand-name drugs based both on the volume bought and on the purchaser's ability to influence market share by systematically favoring one brand-name drug over another. For that reason, one would expect retail pharmacies to pay higher average prices than other purchasers (such as hospitals, long-term care facilities, and health maintenance organizations) because they have less ability to promote such brand-name substitution. (As noted earlier, substituting one therapeutically similar brand-name drug for another requires getting the doctor's consent--something that pharmacists in a hurry do not always have time to do.) If pharmacies do pay higher prices, that may be evidence that some managed care techniques, such as the use of formularies, help other types of purchasers obtain discounts from manufacturers.⁽³⁷⁾ Pharmaceutical benefit management companies, for example, receive rebates from manufacturers precisely because they apply a formulary to a broad patient base, which a retail pharmacy itself generally cannot do.

Types of Discounts

Manufacturers' discounts on brand-name drugs take a variety of forms. Purchasers that buy directly from manufacturers can simply negotiate a lower purchase price. Three-quarters of prescription drugs are bought indirectly, however, through wholesalers. But that does not prevent the purchaser from obtaining a lower price. Manufacturers frequently pay rebates directly to such purchasers based on the volume of drugs they use over a period of time. A demonstrated ability to switch patients to a particular company's drug, evidenced by an increase in the volume used by a purchaser's patient base, may be rewarded with a higher rebate. Some contracts between PBMs and drug companies have been designed in that manner.⁽³⁸⁾

Another important form of discounting involves the wholesaler. Together, manufacturers and wholesalers have developed a computerized system whereby the wholesaler learns of the discounted price negotiated between a manufacturer and a particular purchaser. The wholesaler delivers the drug at the discounted price, informs the manufacturer of the discounted delivery, and then is reimbursed by the manufacturer electronically.(39) Such discounts handled through a wholesaler are generally known as charge-backs (although that term is sometimes used to encompass other types of discounts as well).⁽⁴⁰⁾

Empirical Evidence on Discounting

Most discounts are negotiated privately between manufacturers and purchasers and do not become public information. CBO was able to obtain limited information from IMS America about the different prices that different types of purchasers paid for some prescription drugs in 1993 and 1994 (see Table 4). The prices paid by pharmacies can be viewed as a proxy for the final price paid by customers who do not have a managed drug benefit or PBM to negotiate rebates from manufacturers. That limited pricing information suggests that customers of retail pharmacies who do not have such a plan are paying the most for brand-name drugs.
 

The price comparison is based on the average invoice prices paid by various kinds of purchasers for 100 top-selling drugs sold largely through pharmacies. (Top-selling drugs that were dispensed primarily in an inpatient setting, such as a hospital, were excluded.) About 85 percent of the revenues from sales of those drugs (excluding sales to mail-order pharmacies) came from retail pharmacies; the other 15 percent came from sales to other types of purchasers.

Those other purchasers paid less, on average, than retail pharmacies for the drugs in question. That finding is consistent with the notion that purchasers are rewarded for their ability to influence the prescription choice of a large patient base. For example, hospitals and clinics paid 9 percent less than retail pharmacies in 1994, and HMOs paid 18 percent less. Federal facilities got the biggest discount, over 40 percent, off the average invoice price paid by retail pharmacies.⁽⁴¹⁾

That comparison is based on invoice prices only, which do not capture rebates and other types of discounts that do not appear on an invoice.⁽⁴²⁾ The size of the average price differences between types of purchasers, and perhaps also the relative ranking of the nonpharmacy purchasers, would change if rebates and all forms of discounts were included. But as long as the excluded rebates and discounts were not larger for retail pharmacies than for the other types of purchasers, on average, then the conclusion drawn from Table 4--that customers of pharmacies without a managed drug benefit pay the highest prices for brand-name drugs--would remain correct. Unfortunately, more complete pricing data are not available.

Rebates to PBMs and Medicaid are also not included in Table 4. Such rebates are an important mechanism for lowering the average prices that manufacturers are paid for prescription drugs bought through retail pharmacies. Since the invoice prices paid by pharmacies do not include the rebates that PBMs and Medicaid receive, Table 4 probably overstates the difference between the average prices that manufacturers earn for drugs channeled through retail pharmacies and the average prices they earn for drugs channeled through other types of purchasers.

Statistical Analysis of Discounts

For another perspective on pricing in the pharmaceutical industry, CBO analyzed data on the "best-price discounts" offered by manufacturers of brand-name drugs in 1994. (Manufacturers reported that information to the federal Health Care Financing Administration as part of the Medicaid rebate program.) The best-price discount equals the percentage difference between a manufacturer's best price (the lowest price it offers any private purchaser in the United States) and the average price it charges for drugs distributed to retail pharmacies. The best price encompasses all forms of discounting, whereas the average price to retail pharmacies generally does not include rebates paid to PBMs or Medicaid (although it does include all forms of discounts that manufacturers give directly to pharmacies).⁽⁴³⁾

The best-price discount alone is not a perfect measure of discounting, because it is not representative of all discounts. It would be preferable from an analytic standpoint to know more about the distribution of different prices paid for a particular brand-name drug and the quantity sold at each price. Such extensive pricing data are not publicly available, however.

Manufacturers are very careful about giving large best-price discounts (more than 15.1 percent of their average price to pharmacies) because, by law, they must give that same discount on all drugs distributed through retail pharmacies that are purchased by Medicaid beneficiaries.⁽⁴⁴⁾ Since Medicaid usually constitutes a larger share of a drug's market than any single private purchaser--13 percent of retail pharmacy sales, on average--such a discount can represent a significant reduction in revenues. The Medicaid rebate program makes it less likely that manufacturers would offer a large best-price discount (over 15.1 percent) to just one private purchaser.⁽⁴⁵⁾

CBO's statistical analysis in fact shows that the Medicaid rebate program, which began in 1991, has discouraged discounting on brand-name drugs. For every increase of 3 percentage points in Medicaid's market share for a particular brand-name drug, the best-price discount falls by 1.3 percentage points. (That result does not apply to prescription drugs used exclusively in an inpatient setting, which are generally not included in Medicaid's rebate program.)

The statistical analysis examines the size of discounts offered on brand-name drugs after adjusting for the effects of the number of brand-name and generic competitors, the therapeutic class of the drug, and its Medicaid market share. (For more details of the analysis, see Appendix B.) The results show that the best-price discount on a brand-name drug is 10 to 14 percentage points greater when therapeutically similar brand-name drugs are available from three or more manufacturers. As more producers of brand-name drugs enter a particular therapeutic class, the size of the best-price discount increases. Similar increases occur when generic competitors enter a market. Those results confirm the theory that the steep discounts on brand-name drugs available to some purchasers are a response to competitive market conditions.⁽⁴⁶⁾
 

Competition Between Brand-Name and Generic Drugs

One of the primary goals of the Hatch-Waxman Act was to increase the availability of lower-cost generic drugs. Since the act became law in 1984, the market share of generic drugs has indeed been rising steadily--although not all of that increase stems from the act. For drugs that come in easily countable units, such as tablets and capsules, the share of generic units sold more than doubled between 1984 and 1996--from 18.6 percent of all drug units sold to 42.6 percent (see Figure 6).⁽⁴⁷⁾
 

Figure 6. Growth in the Market Share of Generic Drugs Since 1984

SOURCE: Pharmaceutical Research and Manufacturers of America, 1997 Industry Profile (Washington, D.C.: PhRMA, March 1997), p. 40, based on data from IMS America.

NOTE: Generic market share is calculated as a percentage of all prescription drugs sold, not just off-patent drugs. These figures are based on countable units, such as tablets or capsules; prescription drugs that come in injectible form are not included.

Those numbers are probably the best publicly available estimate documenting the rise in generic market share since 1984. However, since countable units do not include injectable drugs and many types of prescription drugs dispensed in liquid form, they are not a perfect measure of average generic market share. Many injectable drugs are dispensed primarily in hospitals and other inpatient settings, so the estimate may underrepresent those channels of distribution. Countable units appear to yield an estimate of generic market share similar to that measured by the number of prescriptions dispensed through retail pharmacies.⁽⁴⁸⁾

The Hatch-Waxman Act encouraged the entry of generic drugs by establishing an abbreviated approval process for generic versions of all nonantibiotic drugs (antibiotics already had such a process). In addition, the act reversed a 1984 court ruling and allowed generic manufacturers to begin the tests required for FDA approval before the patent on the innovator drug they were copying had expired. Those changes both increased the probability that a generic copy would become available after patent expiration and reduced the average delay between patent expiration and generic entry from more than three years to less than three months.

As generic drugs are substituted for their more expensive brand-name counterparts, the average price of a prescription falls. In CBO's retail pharmacy data set, the average retail prescription price for a brand-name drug with generic substitutes was $37 in 1994. However, including prescriptions that were filled with a generic drug, the average prescription price for a multiple-source drug was only $26. Thus, generic substitution lowered the average cost for a multiple-source prescription by $11. That result is only a rough estimate, however, since prescriptions may somewhat misrepresent the relative quantities of brand-name and generic drugs sold. For example, if generic drugs tend to have more pills dispensed per prescription than their brand-name counterparts, that estimate would understate the degree to which generic substitution reduces the average cost of a prescription. If generic drugs tend to have fewer pills dispensed, the reverse would be true.

Effect of Generic Entry on Sales

For many innovator drugs whose patents have recently expired, generic copies quickly gain a large share of the market. CBO's retail pharmacy data set includes 21 innovator drugs whose first generic competitors entered the market between 1991 and 1993. During the first full calendar year in which those 21 drugs faced generic competition, generics already accounted for an average of 44 percent of prescriptions dispensed through pharmacies.⁽⁴⁹⁾ Generics also cost one-fourth less than the brand-name drugs, on average, at retail prices. For seven of those drugs (Anaprox, Feldene, Lopid, Naprosyn, Pamelor, Tavist, and Xanax), generics had gained 65 percent or more of the innovator's market by 1994. For all but two of the 21 drugs, generic entry occurred within one year of patent expiration, and in many cases within three months.⁽⁵⁰⁾

Other studies examining the size of the generic market after patent expiration have yielded slightly different results. Those appear to be attributable to differences in the sample of drugs studied as well as to small differences in method. A study by Grabowski and Vernon found that 11 drugs whose patents expired between 1989 and 1992 had an average generic market share (measured by quantity sold) of 50 percent in the first year after generic entry, and eight drugs whose patents expired in the 1986-1987 period had an average generic market share of 38 percent.⁽⁵¹⁾ The study also found that the wholesale price of generic drugs was about half that of brand-name drugs in the first year after generic entry.

Grabowski and Vernon's average generic market share for the 1989-1992 period is higher than that measured by CBO for the 1991-1993 period in part because CBO included the quantity sold of all dosage forms of the brand-name drug, even those for which generic entry had not occurred, when calculating the percentage of total prescriptions filled with a generic drug. That method takes account of the option that brand-name manufacturers have to introduce a new dosage form (such as an extended-release capsule) just as a drug's patent is about to expire, so as to benefit from a three-year exclusivity period on that dosage form. Occasionally, manufacturers can even get a separate patent on a new dosage form. Of the 21 brand-name drugs that CBO analyzed, four had an advanced dosage form (Sinemet CR, Cardizem CD, Toprol XL, and Procardia XL) that was not yet available from generic manufacturers.

The Congress's former Office of Technology Assessment (OTA) also studied the erosion of brand-name drug sales after patent expiration.⁽⁵²⁾ In OTA's study of 35 brand-name drugs that lost patent protection between 1984 and 1987, sales volume for those drugs was 43 percent lower three years after patent expiration. Part of the reason it took that long for brand-name sales to erode by so much was a longer delay between patent expiration and generic entry during the period that OTA examined. For more than half of the 1984-1987 period, generic manufacturers could not have begun the abbreviated drug-approval process far enough in advance to enter the market soon after patent expiration. Also, that study differed from CBO's analysis because it focused on the decline in brand-name sales following patent expiration rather than explicitly on generic market share. Actual generic market share measured in volume may have been greater than 43 percent if the total quantity of the drugs demanded rose because generic drugs were cheaper. Or generic market share may have been smaller if competition from similar brand-name drugs was also eroding innovators' sales. OTA's estimates also differed from CBO's in that its measurements were based on the date of patent expiration rather than the date of generic entry.

Before 1984 and the Hatch-Waxman Act, competition from generic drugs in terms of price and market share was limited primarily to antibiotics.⁽⁵³⁾ In 29 cases other than antibiotics in which top-selling brand-name drugs had generic copies available, generic market share averaged just 12.7 percent of prescriptions dispensed through retail pharmacies in 1980.⁽⁵⁴⁾ The probability of generic entry was also much lower before 1984. Excluding antibiotics and drugs approved before 1962 (for which an abbreviated generic-drug-approval process existed), only 18 out of 52 top-selling drugs with expired patents had generic versions available.⁽⁵⁵⁾ Clearly, the lengthy FDA approval process at that time hampered the generic drug industry.

Effect of Generic Entry on Brand-Name Prices

Those consumers who are more sensitive to price, or who are covered by health plans that encourage generic substitution, are more likely to buy a generic drug when it becomes available. As the more price-sensitive consumers switch to the generic version, demand for the original brand-name drug declines and may become less sensitive to price. If that happens, the price of the brand-name drug could theoretically rise more quickly over time than it would have without generic competition.⁽⁵⁶⁾

A number of empirical studies have found that the prices of brand-name drugs continue to rise faster than inflation after generic entry (see Box 4 for details). One study also found that brand-name prices increase by about 1 percent with each new generic competitor. At the same time, CBO's analysis shows that discounts on brand-name drugs tend to increase after generic entry, something not fully captured in the invoice prices on which the other empirical studies are based. CBO found that the best-price discount is 10 to 17 percentage points greater when two or more generic manufacturers produce copies of the brand-name drug (see Appendix B). Taken together, the implication of those results is that prices of brand-name drugs do rise faster than inflation for many final purchasers after generic entry, but some purchasers pay less for those drugs after generic entry.
 

CBO examined the prices that manufacturers charged for 34 brand-name drugs distributed to retail pharmacies that first saw generic competition after 1991. It found that those brand-name prices continued to increase faster than inflation after generic entry, perhaps as much as they would have otherwise.⁽⁵⁷⁾ That result affects primarily third-party payers that do not manage their outpatient drug benefits and consumers who have no insurance (but who still purchased the brand-name drug). Other types of purchasers, such as Medicaid and PBMs, get rebates from manufacturers that are not captured in the prices charged to pharmacies.

For 34 drugs that experienced generic competition for the first time after 1991, the average price increase between 1991 and 1994 was 22 percent. By comparison, average prices for brand-name drugs that faced no generic competition rose by 24.5 percent over that period. And the prices of brand-name drugs that had already faced generic competition by 1991 grew by 22.4 percent during the same period. (Apart from any effect of generic competition, that price increase for multiple-source drugs could be lower because many of the drugs are older ones that have been surpassed by newer treatments.) The differences in the rate of price increase among those three groups of brand-name drugs are small and consistent with the notion that generic competition does not have a large effect on brand-name prices for many purchasers.

Effect of Generic Competition on Total Costs for Prescription Drugs

Because generic drugs are priced much lower than their brand-name counterparts, they are a source of substantial savings. According to CBO's data on retail pharmacy sales, the average retail price of a prescription for a generic drug in 1994 was $17.40 (see Table 1). Multiple-source brand-name drugs were twice as expensive--averaging $37.40 per prescription.

CBO estimates that if each generic prescription had been dispensed at the corresponding brand-name price, purchasers of prescription drugs through retail pharmacies would have spent roughly $8 billion to $10 billion more in 1994. Those figures were calculated as follows: CBO assumed that all of the generic prescriptions dispensed in 1994 would have been filled with a higher-priced brand-name drug if the generic drug was not available.⁽⁵⁸⁾ Then the price difference between the innovator and generic formulations of a given drug was multiplied by the number of generic prescriptions dispensed for that drug. Adding together the results of those calculations for all of the multiple-source drugs in the retail pharmacy data set yielded an estimate of $7 billion in direct savings from retail purchases of generic drugs in the data set.⁽⁵⁹⁾

The sales data cover only 70 percent of the retail pharmacy market, however, although they may cover more than 70 percent of generic drug sales through retail pharmacies since they include nearly all of the 200 top-selling drugs that are dispensed primarily through pharmacies. Assuming that the data set encompasses 70 percent to 90 percent of total generic sales, then savings from all retail purchases of generic drugs through pharmacies would total approximately $8 billion to $10 billion in 1994. Of course, retail pharmacies are not the only sellers of prescription drugs. Since other channels (including hospitals, clinics, and mail-order pharmacies) distribute around 40 percent of prescription drugs, the total savings from generic substitution through all channels were most likely even greater than that amount.

That calculation entails a variety of assumptions and caveats. First, it assumes that the quantity of prescriptions filled for a particular multiple-source drug does not increase because a lower-priced generic has become available. If the number of prescriptions did increase, the calculation would overstate the savings from generic entry. However, limited statistical evidence supports the assumption that the quantity sold does not change. A study by Caves, Whinston, and Hurwitz found that the total amount sold of a drug in both generic and brand-name forms did not increase after generic entry.⁽⁶⁰⁾

Second, the calculation is a rough one because the price per prescription, from which it is derived, does not account for possible systematic differences between the size of brand-name and generic prescriptions. The calculation would be more accurate--though much more cumbersome--if the unit of measure was the cost of an average daily dose. But even that measure contains problems, because the average daily dose can vary among people and among the different medical conditions that a drug is used for. Without the ability to use a better measure, the calculation relies on prescriptions as the unit of quantity to obtain a rough estimate of the savings from generic substitution.

Finally, the calculation does not include any rebates that manufacturers pay to PBMs or other purchasers of prescription drugs through retail pharmacies. Excluding those rebates leads to an overestimation of the savings from generic substitution at retail pharmacies. That overestimate could be as much as roughly $500 million, assuming that manufacturers give rebates on multiple-source brand-name drugs (to PBMs and other third-party payers that manage their outpatient drug benefits) to the same extent that they do on brand-name drugs still under patent.⁽⁶¹⁾
 

Competition Among Generic Drugs

The expiration of an innovator drug's patent frequently prompts more than one generic copy to enter the market. Since most generic competitors sell their copy under the same chemical name, there is little apparent difference between their products. Economic theory suggests that differences between products dampen price competition, so when products are roughly identical, price competition can be intense. Hence, as more generic manufacturers enter the market, they should face increased pressure to lower prices in order to maintain market share.

Tabulations of average retail prescription prices in 1994 show that the average price of a generic drug does decline as the number of manufacturers and distributors of that drug increases (see Table 5). For example, the average prescription price of a generic drug with one to five manufacturers ($23.40) is more than that of a drug with 16 to 20 manufacturers ($19.90). CBO's retail pharmacy data set covers 112 innovator drugs that in 1994 were also available in generic forms sold under their chemical name. Comparing the average generic prescription price with the average innovator price for the same drug also shows prices falling as the number of generic manufacturers rises. When one to 10 generic manufacturers are in the market, the generic retail prescription price averages 61 percent of the brand-name price. When 11 to 24 generic manufacturers are in the market, the generic retail price averages less than half of the brand-name price.
 

Other studies have also concluded that prices of generic drugs decline in response to increased generic competition. Economist Richard Caves and colleagues found that as the number of generic manufacturers increased from one to 10, the average generic price fell from 60 percent to just 34 percent of the brand-name price. With 20 manufacturers, the generic price was only 20 percent of the brand-name price.⁽⁶²⁾ Since generic prices tend to fall as the number of producers rises, generic manufacturers are most profitable when they are one of the first to enter a market.

Market Concentration in the Generic Drug Industry

Overall, the generic drug market is not particularly concentrated. Mylan and Geneva, the largest generic firms in 1994, accounted for 16 percent and 12 percent, respectively, of all generic sales in the retail pharmacy data set. Most generic firms had just 1 percent to 5 percent of total generic sales.

The markets for individual multiple-source drugs, by contrast, are much more concentrated. For 94 of 110 multiple-source drugs in the retail pharmacy data set, the top two generic firms were responsible for more than half of generic sales. And for 57 of those drugs, the single top generic firm accounted for more than half of generic sales.

Leading generic firms may lower their price when new competitors enter the market so as to maintain their dominant position. That would explain how the average generic price falls as the number of manufacturers rises, but sales of many generic drugs remain dominated by one or two companies. Still, Grabowski and Vernon found that in only half of the 18 markets they examined, the lowest-priced generic manufacturer had the largest market share.⁽⁶³⁾ Factors other than price, such as being the first to enter a market, probably also play a role in determining a generic manufacturer's market share. And one recent study found that generic manufacturers are more likely to enter markets where they have some experience with a drug's dosage form, therapy, or active ingredient.⁽⁶⁴⁾

Links Between Generic and Brand-Name Manufacturers

Although the same company rarely produces both a brand-name drug and its generic copy, some generic manufacturers are subsidiaries of brand-name firms. In 1994, eight of the 15 largest generic companies in the retail pharmacy data set were owned by innovator firms.⁽⁶⁵⁾ Those generic subsidiaries were responsible for 46 percent of total generic sales in the data set.

Today, the proportion of generic drugs produced by subsidiaries of innovator firms is probably somewhat smaller than in 1994 because several brand-name manufacturers have left the generic drug business. For example, three of the eight larger generic firms owned by a brand-name company (Rugby, Hamilton, and Warner-Chilcott) have been sold or disbanded in recent years.⁽⁶⁶⁾ Some of those brand-name companies experimented with producing generic copies of their own drugs in the early 1990s and found that it was not very profitable. For example, generic manufacturer Hamilton offered copies of the brand-name drugs Anaprox and Naprosyn produced by its parent company, Syntex. During the first calendar year after patent expiration, the average generic price quickly dropped, and Syntex lost 70 percent of its market for those two drugs to generic competition.⁽⁶⁷⁾ A few of the brand-name companies that tried to get further into the generic business in the early 1990s, including Hoechst Marion Roussel and Merck, have recently sold generic subsidiaries.⁽⁶⁸⁾

Nevertheless, brand-name companies that have long held generic subsidiaries remain committed to their generic business. Today, at least 13 manufacturers of innovator drugs have a generic subsidiary or division (see Table 6). One of the largest generic firms, Geneva Pharmaceuticals, is a subsidiary of Novartis (a company formed by the merger of Ciba-Geigy and Sandoz).
 

Most generic subsidiaries do not produce copies of their parent company's drugs. Out of 112 multiple-source drugs in the retail pharmacy data set, only 13 had a generic subsidiary of the brand-name manufacturer selling more than 10 percent of the prescriptions dispensed through retail pharmacies. In general, the incentives to lower price in order to gain market share are the same for all generic manufacturers, whether or not they are the subsidiary of an innovator firm. But an important exception occurs when the generic subsidiary produces a copy of the parent company's innovator drug. Though infrequent, in such cases the subsidiary may have less incentive to lower price than other generic producers because it does not want to take more sales away from the parent company's drug. And when the generic subsidiary does lower price dramatically, the innovator firm suffers.
 

Conclusions

Changes to the approval process for generic drugs made by the Hatch-Waxman Act, combined with the changes in demand for generic drugs discussed in Chapter 2, have prompted a dramatic rise in generic competition since 1984. That increased competition has helped hold down the average price of a multiple-source prescription drug by encouraging the substitution of lower-priced generic drugs for brand-name ones. In 1994, such substitution saved final purchasers of prescription drugs through retail pharmacies roughly $8 billion to $10 billion (at retail prices).

Manufacturers of generic drugs, who sell nearly identical versions of the same product, compete more intensely on the basis of price than do manufacturers of innovator drugs, who compete more on the basis of quality and other differences between products. Average list and invoice prices of brand-name drugs do not typically fall after generic competitors enter the market. On a selective basis, however, manufacturers of brand-name drugs do offer discounts and rebates to some purchasers, and those discounts tend to be larger when generic versions of the drug are available. The data necessary to determine what volume of purchases is sold at a substantial discount do not exist. The industry group Pharmaceutical Research and Manufacturers of America estimates that discounts saved purchasers $5.3 billion in 1994, $3.5 billion of which went to non-Medicaid purchasers.⁽⁶⁹⁾ (That $3.5 billion represented over 5 percent of the value of non-Medicaid prescription drug sales.)

The extent to which brand-name drugs compete through price is difficult to assess. Limited empirical evidence suggests that competition between similar brand-name drugs causes their prices to rise more slowly over time than would otherwise be the case. However, evidence also suggests that the prices of me-too drugs increase much more rapidly over time than the price of the breakthrough drug. Much of that analysis is based on list prices or average invoice prices, which do not include many charge-backs and rebates.

Clearly, some price competition is occurring, particularly in the segment of the market that can negotiate discounts when several similar brand-name drugs are available. As Chapter 2 noted, that segment of the market is growing with the emergence of PBMs and the proliferation of other managed care techniques. Still, since the size of discounts and the quantity of drugs sold at a discount are not known, it is difficult to assess the extent of competition brought about through discounting.

1. That figure represents the after-tax cost of R&D and was calculated as follows: for drugs developed between 1970 and 1982, manufacturers' out-of-pocket costs were about $100 million per drug, after averaging in the costs of clinical failures. Accounting for the opportunity cost of capital (or the time value of money) nearly triples those costs. But since R&D investments are expensed for tax purposes (because a dollar invested in R&D is a dollar on which corporate profit taxes are not paid), the after-tax cost comes to about $200 million at a marginal tax rate of 35 percent. See Office of Technology Assessment, Pharmaceutical R&D: Costs, Risks and Rewards (February 1993); and Henry G. Grabowski and John M. Vernon, "Returns to R&D on New Drug Introductions in the 1980s," Journal of Health Economics, vol. 13, no. 4 (December 1994), pp. 383-406. Both of those studies rely on Joseph A. DiMasi and others, "Cost of Innovation in the Pharmaceutical Industry," Journal of Health Economics, vol. 10, no. 2 (July 1991), pp. 107-142.

2. For a discussion about how changes in technology have affected the R&D process, see Geoffrey Carr, "A Survey of the Pharmaceutical Industry," Economist, February 21, 1998. As one example, computer programs are being developed that can help predict whether a clinical trial is likely to work before it is undertaken.

3. See Grabowksi and Vernon, "Returns to R&D"; and Office of Technology Assessment, Pharmaceutical R&D. Those measures account for the cost of capital, so the returns are beyond the amount necessary to adequately compensate investors for their investment in drug development.

4. Joseph A. DiMasi, Mark A. Seibring, and Louis Lasagna, "New Drug Development in the United States from 1963 to 1992," Clinical Pharmacology and Therapeutics, vol. 55, no. 6 (June 1994), pp. 609-622.

5. See DiMasi and others, "Cost of Innovation in the Pharmaceutical Industry"; and Blanchard Randall, Drug Regulation: Historical Overview and Current Reform Proposals, CRS Report for Congress 95-962 SPR (Congressional Research Service, September 11, 1995), pp. 7-8.

6. David A. Kessler and Karyn I. Feiden, "Faster Evaluation of Vital Drugs," Scientific American (March 1995).

7. Ibid., p. 50. For an explanation of the need for a large clinical trial to demonstrate that a drug is as safe as, and more effective than, existing treatments, see F.M. Scherer, Industry Structure, Strategy, and Public Policy (New York: Harper Collins College Publishers, 1996), pp. 353-355.

8. Section 736 of the Federal Food, Drug, and Cosmetic Act of 1938, as amended, 21 U.S.C. 379(h).

9. See Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research, Center for Drug Evaluation and Research Fact Book, 1997 (May 1997), available at http://www.fda.gov/cder/reports/cderfact.pdf. Note that less than half of all new drug applications include a chemical entity never before approved.

10. Federal Food, Drug, and Cosmetic Act of 1938, as amended, 21 U.S.C. 355(c)(1).

11. See Department of Health and Human Services, Food and Drug Administration, Fourth Annual Performance Report, Prescription Drug User Fee Act of 1992 (December 1, 1996), available at http://www.fda.gov/ope/96pdufa.htm.

12. See DiMasi, Seibring, and Lasagna, "New Drug Development in the United States from 1963 to 1992."

13. Food and Drug Administration Modernization Act of 1997, 21 U.S.C. 393.

14. Food and Drug Administration Modernization Act of 1997, 21 U.S.C. 355(b).

15. According to Office of Technology Assessment, Pharmaceutical R&D, and Grabowski and Vernon, "Returns to R&D," the average present discounted value of the profit stream from marketing a new drug over its product life is $210 million to $230 million in 1990 dollars. Thus, at an interest rate of 10 percent, adding a year to that product life by speeding up market introduction could raise returns by $21 million to $23 million.

16. Ibid. Those number are also at a present discounted value in 1990 dollars.

17. Z. John Lu and William S. Comanor, Strategic Pricing of New Pharmaceuticals, Working Paper 96-1 (Los Angeles: University of California School of Public Health, Research Program in Pharmaceutical Economics and Policy, October 9, 1996), p. 1.

18. The 13 therapeutic classes were H2 antagonists, beta-blockers, ace inhibitors, cholesterol reducers, serotonin reuptake inhibitors (antidepressants), 5-HT3 receptor antagonists (antinauseants), cephalosporins (1st, 2nd, and 3rd generations), growth hormones, calcium channel blockers, loop diuretics, and benzodiazepines (tranquilizers). Those classes were defined by a mechanism of action clearly distinguished in Facts and Comparisons, Drug Facts and Comparisons (St. Louis, Mo.: Facts and Comparisons, 1995).

19. See Ernst Berndt and others, The Roles of Marketing, Product Quality and Price Competition in the Growth and Composition of the U.S. Anti-Ulcer Drug Industry, Working Paper No. 4904 (Cambridge, Mass.: National Bureau of Economic Research, October 1994).

20. Economists refer to this as the point at which incremental, or marginal, revenue from selling another unit of the drug is equal to the cost of producing another unit. To keep Figure 4 simple, the cost of producing another unit is assumed to be the same no matter how much is produced (therefore, unit production costs are represented by a horizontal line).

21. The list price, called the average wholesale price, or AWP, is published annually in Medical Economics Company, Red Book (Montvale, N.J.: Medical Economics Company).

22. Prices were obtained from the 1980 to 1994 editions of the Red Book. The five therapeutic classes were H2 antagonists, cholesterol reducers (specifically HMG-CoA reductase inhibitors), antidepressants (specifically serotonin reuptake inhibitors), fluoroquinolone anti-infectives, and alpha-blockers, as listed in Facts and Comparisons, Drug Facts and Comparisons.

23. In one of those four cases, the entry price of the me-too drug exceeded that of the breakthrough drug. In the other three, the breakthrough drug's price was not reduced even though the me-too drugs with which it competed were available at a lower price.

24. Lu and Comanor, Strategic Pricing of New Pharmaceuticals.

25. Economists have analyzed this phenomenon using an "experience goods" or "switching costs" model. See F.M. Scherer and David Ross, Industrial Market Structure and Economic Performance (Boston: Houghton Mifflin, 1990), pp. 588-589.

26. Scherer, Industry Structure, Strategy, and Public Policy, p. 351.

27. The four classes were gout therapies, nonbarbiturate sedatives, oral diabetic therapies, and minor tranquilizers. See Scott Stern, "Product Demand in Pharmaceutical Markets" (draft, Stanford University, Department of Economics, November 21, 1994; the draft was updated in 1996 at MIT's Sloan School of Management).

28. See, for example, W. Duncan Reekie, "Price and Quality Competition in Drug Markets: Evidence from the United States and the Netherlands," in Robert B. Helms, ed., Drugs and Health: Economic Issues and Policy Objectives (Washington, D.C.: American Enterprise Institute, 1981).

29. Richard E. Caves, Michael D. Whinston, and Mark A. Hurwitz, "Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry," Brookings Papers on Economic Activity: Microeconomics (1991), pp. 11-12; and Mark A. Hurwitz and Richard E. Caves, "Persuasion or Information? Promotion and the Share of Brand-Name and Generic Pharmaceuticals," Journal of Law and Economics, vol. 31 (October 1988), p. 302.

30. Economists would also say that economies of scope are important. Economies of scope occur when the production or advertising of more than one product lowers the average cost of those expenditures for all products.

31. General Accounting Office, FDA Drug Approval Review Time Has Decreased in Recent Years, GAO/PEMD-96-1 (October 1995), p. 5. "Experienced sponsors" submitted nine or more NDAs between 1987 and 1992, whereas "inexperienced sponsors" submitted four or fewer NDAs and had no affiliation with more experienced sponsors.

32. Large firms may also have advantages in financial markets, overcoming problems of adverse selection and moral hazard to obtain funding more easily. And they can more easily fund a drug's development out of their profits from sales.

33. Based on U.S. sales reported by Med Ad News (September 1995), p. 34.

34. Thirteen of the therapeutic classes contained just one to three innovator drugs. In five of those 13 classes, the top three innovator drugs had less than 63 percent of the market because of generic competition.

35. For a general discussion of price discrimination, see Jean Tirole, The Theory of Industrial Organization (Cambridge: MIT Press, 1989), Chapter 3. For a discussion of the legal and economic issues surrounding pricing practices in the pharmaceutical industry, see "Symposium on the Brand Name Prescription Drug Antitrust Litigation," International Journal of the Economics of Business, vol. 4, no. 3 (November 1997).

36. See Tirole, The Theory of Industrial Organization, pp. 137-139.

37. For a further discussion of this issue, see Kenneth G. Elzinga and David E. Mills, "The Distribution and Pricing of Prescription Drugs," International Journal of the Economics of Business, vol. 4, no. 3 (November 1997), pp. 289-292.

38. See, for example, "PCS Rebates from Pfizer on Seven Products Totaled over $10 Million in First 21 Months of 1994-1998 Contract," The Pink Sheet, F-D-C Reports, June 10, 1996, p. 16.

39. For a discussion of that system, see F.M. Scherer, "How U.S. Antitrust Can Go Astray: The Brand Name Prescription Drug Litigation," International Journal of the Economics of Business, vol. 4, no. 3 (November 1997), p. 248.

40. For example, hospitals and hospital buying groups sometimes refer to the rebates paid directly to them by manufacturers for drugs bought through wholesalers as charge-backs, even though the wholesalers have no knowledge of them. See Caves, Whinston, and Hurwitz, "Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry," p. 32.

41. Note that the prices paid by federal agencies--such as the Department of Veterans Affairs, the Defense Department, the Indian Health Service, and the Public Health Service, as well as state pharmaceutical assistance programs--are not affected by the best-price provision in the Medicaid rebate program, which discourages discounting. That exclusion was made permanent by the Veterans Health Care Act of 1992. For more information, see Congressional Budget Office, How the Medicaid Rebate on Prescription Drugs Affects Pricing in the Pharmaceutical Industry, CBO Paper (January 1996); and General Accounting Office, Drug Prices: Effects of Opening Federal Supply Schedule for Pharmaceuticals Are Uncertain, GAO/HEHS-97-60 (June 1997).

42. Invoice prices generally incorporate discounts granted through a charge-back system with wholesalers.

43. For more detailed information on the calculation of those prices, see the Medicaid rebate agreement signed by manufacturers (available at http://cms.hhs.gov/medicaid/default.asp). The calculation of the average price that manufacturers charge for drugs distributed to retail pharmacies includes sales and discounts to mail-order pharmacies.

44. Manufacturers pay at least a flat rebate of 15.1 percent of the average manufacturer price for drugs they distribute through retail pharmacies that are purchased by Medicaid beneficiaries. The rebate percentage is equal to the best-price discount only when that discount exceeds 15.1 percent.

45. See Congressional Budget Office, How the Medicaid Rebate on Prescription Drugs Affects Pricing in the Pharmaceutical Industry, pp. 22-25.

46. CBO's 1996 paper on the Medicaid rebate program also found that the largest discounts were significantly higher for multiple-source drugs than for single-source drugs. In 1991, the largest discounts offered on multiple-source innovator drugs averaged 50 percent off the price to pharmacies, compared with 35 percent off for single-source drugs; see Congressional Budget Office, How the Medicaid Rebate on Prescription Drugs Affects Pricing in the Pharmaceutical Industry. See also Fiona Scott Morton, "The Strategic Response by Pharmaceutical Firms to the Medicaid Most-Favored-Customer Rules," RAND Journal of Economics, vol. 28, no. 2 (Summer 1997), pp. 269-290.

47. Those figures come from IMS America and are published in Pharmaceutical Research and Manufacturers of America, 1997 Industry Profile, p. 40. The publication gave the generic market share in 1996 as 41.6 percent. The corrected 1996 figure came from a personal communication from Paul Wilson, Vice President of Statistical Services, IMS America, on February 27, 1998.

48. "Approximately 57 percent of all prescriptions paid for by managed care are still filled with branded products--a virtually identical ratio to the overall market," implying a generic market share of about 43 percent for the retail pharmacy market; see IMS America, "IMS Says Managed Care Drove Unprecedented Growth in Pharmaceuticals in 1996" (press release, April 14, 1997).

49. The 44 percent average is weighted by sales revenues of the innovator drugs. The unweighted average is 42.8 percent.

50. The two drugs for which generic entry took more than a year after patent expiration had retail pharmacy sales of about $130 million in 1991.

51. Henry Grabowski and John Vernon, "Longer Patents for Increased Generic Competition in the U.S.: The Hatch-Waxman Act After One Decade," PharmacoEconomics (1996).

52. Office of Technology Assessment, Pharmaceutical R&D, Table F-3, p. 297.

53. See Federal Trade Commission, Bureau of Consumer Protection, Drug Product Selection (1979), p. 46.

54. See Appendix C for details.

55. Those drugs were all in the top 200 drugs in the United States, rated by sales. 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. 195-198.

56. Frank and Salkever have developed a theoretical model that formally captures this phenomenon, showing that it may be profitable for the manufacturer of the innovator drug to raise its price after generic entry; see Richard G. Frank and David S. Salkever, "Pricing, Patent Loss and the Market for Pharmaceuticals," Southern Economic Journal (October 1992), pp. 165-179.

57. The analysis was based on the average price that manufacturers charged for brand-name drugs sold to the retail pharmacy class of trade, as reported by manufacturers to the Health Care Financing Administration as part of the Medicaid rebate program. Those prices, which include all discounts and rebates to retail pharmacies, were matched to the drugs in the retail pharmacy data set to determine whether a generic substitute existed. (For more details on the pricing data, see Appendix A.)

58. Technically, the calculation assumed that demand is perfectly price inelastic--that is, the lower price of generic drugs does not induce more prescriptions to be filled than if the cheaper generic version did not exist. To the extent that people fill prescriptions they would have left unfilled if a cheap generic version was not available, the estimate somewhat overstates the savings from generic substitution. And to the extent that some consumers substitute the generic for a therapeutically similar (but chemically different) brand-name drug that is still under patent, savings from generic substitution exist but the calculation estimates them based on the wrong brand-name price. That may or may not lead to a small overstatement of the total savings.

59. Those savings were calculated only for tablet and capsule dosage forms, which constitute 91 percent of the value of generic sales in the retail pharmacy data set. Those dosage forms yield a more reliable average price per prescription, which forms the basis of the calculation.

60. Caves, Whinston, and Hurwitz, "Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry."

61. Discounts and rebates to private purchasers in 1994 totaled $3,456 million (not including Medicaid rebates), according to information that the Pharmaceutical Research and Manufacturers of America provided to CBO on April 28, 1997. Pharmacies distribute 60 percent of prescription drugs, but only rebates to third-party payers, not the discounts to pharmacies themselves, should be counted. Assuming that 40 percent of the discounts and rebates went to PBMs and other purchasers that manage their outpatient drug benefits (a very generous amount), that leaves $1,382 million. Since multiple-source brand-name drugs represent about 33 percent of the value of all brand-name drugs sold through retail pharmacies, taking 33 percent of that leaves $455 million.

62. Caves, Whinston, and Hurwitz, "Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry," p. 36, Table 9. Their study actually counted the number of approved abbreviated new drug applications, which a generic manufacturer is required to obtain from the FDA, rather than the number of manufacturers and distributors.

63. Henry Grabowski and John Vernon, "Brand Loyalty, Entry, and Price Competition in Pharmaceuticals After the 1984 Drug Act," Journal of Law and Economics (October 1992), p. 345. CBO's retail pharmacy data are in retail prices, so they cannot be used to compare the prices charged by different generic manufacturers.

64. Fiona Scott Morton, Entry Decisions in the Generic Pharmaceutical Industry, Working Paper No. 6190 (Cambridge, Mass.: National Bureau of Economic Research, September 1997).

65. All 15 companies had annual sales of over $100 million for the drugs in the retail pharmacy data set in 1994.

66. Rugby, which was owned by Hoechst Marion Roussel, was sold to Watson, a generic drug company. Hamilton, a subsidiary of Syntex, was disbanded when Syntex was acquired by Roche in 1995. And Warner-Chilcott was sold by Warner-Lambert to Nalé Laboratories.

67. Based on CBO's retail pharmacy data set. Also see Catherine Yang, "The Drugmakers vs. the Trustbusters," Business Week, September 5, 1994, p. 67.

68. Milt Freudenheim, "Cleaning Out the Medicine Cabinet," New York Times, September 11, 1997, p. D1. Hoechst Marion Roussel sold Rugby in 1997 but still owns two smaller generic subsidiaries.

69. The group's $5.3 billion estimate is based on reporting from its member companies. In 1994, manufacturers paid states $1.8 billion under the Medicaid rebate program, leaving a net value of $3.5 billion in discounts to non-Medicaid purchasers.

---