PERSUASION OR INFORMATION?
`
`PROMOTION AND THE SHARES OF
`
`BRAND NAME AND GENERIC
`
`PHARMACEUTICALS*
`
`and
`
`MARK A. HURWITZ
`RICHARD E. CAVES
`Carpomre Decisions. Inc.
`Harvard University
`Ecostomsrs have vigorously debated whether advertising and other
`messages supplied by sellers to buyers represent the efficient provision of
`information or the exploitation of buyers‘ imperfect access to it. Many
`economists now agree that each view commands some truth. Advertising
`should convey information efficiently where the buyer can easily verify it.
`But it may engender inefficient rent—seeking outlays by producers able to
`hamper buyers‘ gaining of information from alternative sources. For ex-
`ample, if buyers sample product information randomly, an incumbent can
`“jam” the channels through which entrants transmit their messages by
`loading the sampled pOpulation with messages of his own.l Or the incum-
`bent‘s messages can reinforce buyers‘ habits so as to reduce their prior
`expectations of the value of trying an alternative brand.2 If sales promo-
`tion is effective (by whatever means) in causing buyers to shift among
`competing products, it becomes a form of rent-seeking outlay by which
`sellers bid for the available customers.3 The problem for empirical re-
`search is to determine the extent to which seller—supplied information
`pursues a rent-seeking goal and thus incurs social costs. Those costs must
`be set against the efficiency advantage of sellers (relative to buyers or
`
`* We are grateful to Dennis Carlton and a referee for suggestions and to Heng-fu Zou for
`research assistance.
`
`' William S. Comanor. The Political Economy of the Pharmaceutical industry. 24 J.
`Econ. Literature 1178 ([986). at [191 and references cited therein.
`2 Richard Schmalensee‘ Product Differentiation Advantages of Pioneering Brands. 72
`Am. Econ. Rev. 349 ([982]. Some evidence on the incidence ofinformative and rent—seeking
`outlays in sales promotion is provided by Richard E. Caves. Information Structures of
`Product Markets. 24 Econ. Inquiry 195 (1986}.
`3 See. for example. Sherwin Rosen, Advertising. Information, and Product Differentia—
`tion. in Issues in Advertising: The Economics of Persuasion l6l, [77—82 (David G. Tuerck
`ed.
`[978).
`[Journal ofLow & Ecoriomics. Vol. XXXI (October 1988)]
`© 1988 by The University of Chicago. All rights reserved. 0022-2186f88f3102-0002$01.50
`
`299
`
`IMMUNOGEN 2288, pg. 1
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`300
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`third parties) as suppliers of product information demanded by buyers—
`as Leffler put it, persuasion or information.‘
`The pharmaceutical industry provides a strategic site for testing hy-
`potheses about
`the scope of rent seeking in manufacturers‘
`sales-
`promotion outlays. Obviously. physicians who prescribe drugs must ac-
`quire extensive information about
`the uses and limitations of new
`pharmaceuticals. The manufacturer who must generate the bulk of this
`information in the course of developing a drug has both the opportunity
`and the incentive to disseminate it efficiently while the drug enjoys patent
`protection. When the patent expires the innovator‘s trademark lives on,
`but competitors may arise to offer the generic drug at discount prices. The
`incumbent may sustain its position against such rivals both by current
`outlays on promotion to jam the entrants’ information channels and by
`anticipatory investments to enlarge the goodwill asset of its trademarked
`brand. Prescribers’ weak incentives for selecting the lowest-price brand
`enhance the payout to such policies. The physician who prescribes a drug
`on the basis of seller—supplied information captures no savings from
`selecting a cheaper generic supplier (these savings seem too small to
`attract patients away from other physicians], so that the price elasticity
`of demand for the branded drug is reduced, and the innovator obtains
`something approximating the advantage of pioneering brands modeled
`by Schmalensee.5
`For a sample of drugs that have gone off—patent and encountered ge-
`neric competition, we address a series of questions about rivalry in sales
`promotion and pricing between the innovators and the generic entrants.
`I.
`Is the entrants‘ market share adversely affected by the innovator’s
`accumulated goodwill asset or by his ongoing sales-promotion outlays?
`2. How effective are the generics1 sales-promotion outlays and price
`discounts in eroding the innovator’s market share?
`3. How is this rivalry affected by the passage of time—the number of
`years in which the innovator enjoyed a monopoly and the number of years
`since generics first invaded the market?
`4. Is the generics1 competition less effective (and are innovators‘ de—
`fenses more effective) in the pharmacy than in the hospital market, where
`incentives to minimize costs operate more strongly?
`the
`in Section I‘
`Before presenting the design in detail we review,
`evidence of the pharmaceuticals market’s susceptibility to rent-seeking
`sales promotion and characterize the resulting opportunity for dynamic
`
`“ Keith B. Leffler. Persuasion or Information? The Economics of Prescription Drug Ad—
`vertising, 24 J. Law & Econ. 4?—48 (1981}.
`5 Schmalensee, supra note 2.
`
`IMMUNOGEN 2288, pg. 2
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROMOTION OF PHARMACEUTICALs
`
`301
`
`optimization by the innovator. Section II outlines the statistical model,
`evaluates the results, and reports tests of several corollary hypotheses.
`Section III relates our findings to recent changes in public policy toward
`pharmaceutical innovators and their competitors.
`
`1. CHARACTERtsths OF THE PHARMACEUTICALS MARKET
`
`We review evidence on the positions of pharmaceutical innovators,
`entrant producers of generic drugs, and health professionals in order to
`show the role of sales-promotion activities in the decisions made by each
`group.
`
`Producers of Branded Phormacemicais
`
`A pharmaceutical firm that acquires a patent on a new ethical drug
`becomes a temporary monopolist who knows when its legal protection
`against entrants will expire. Theoretical analyses of such monopolists‘
`behavior have stressed their scope for maximizing wealth by building a
`goodWill asset while entry is precluded and by responding optimally to
`entry When it occurs. If the monopoly holds no durable goodwill asset, its
`position When legal protection lapses becomes no different from that of
`new entrants to the market (scale economies and sunk costs permitting).
`If the monopoly holds a durable but wasting goodwill asset, or if the rate
`of newcomers’ entry is constrained, the monopoly enjoys strategic op-
`tions in the postpatent period that boil down to ceding market share to
`entrants at a rate that maximizes the terminal value of its goodwill asset.°
`During the period of our study, by the time a patented drug was ap-
`proved by the U.S. Food and Drug Administration for marketing.
`its
`developer typically had about half of its seventeen-year patent protection
`left as a period of legal monopoly? However, the developer gains perma-
`nent monopoly of a trademark name that becomes the vehicle for a dur-
`able goodwill asset. Thus, any given drug (like T. S. Eliot‘s cat) has three
`different names, One, the chemical name, describes the product‘s molec—
`
`° This large literature includes both theoretical models such as Darius W. Gaskins, IL,
`Dynamic Limit Pricing: Optimal Pricing under Threat of Entry. 3 J. Ec0n. Theory 306
`([971]; and empirical tests such as Robert T. Masson & Joseph Shaanan1 Stochastic—
`Dynamic Limit Pricing: An Empirical Test, 64 Rev. Econ. Stat. 413 (1982}.
`’ Mean-effective patent life had declined from about sixteen years in the mid-19605 to
`seven to nine years after [978. See Richard A. Spivey & A. Gene Trimble, Effect ofthe Drug
`Price Competition Acton Market Exclusivity of New Drugs: A Simulation, 20 Drug infor-
`mation J. 27(1986]. Although drugs on patent are sometimes licensed to another company or
`soldjointly by two companies as part ofa dual marketing arrangement, most drug products
`sold by their innovators do not encounter competition until the patent expiration date.
`
`IMMUNOGEN 2288, pg. 3
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`302
`
`THE JOURNAL OF LAW AND economics
`
`ular structure to scientists. The generic name is a shorter, simpler version
`of the chemical name and is not protected by a trademark. The brand
`name assigned by the developer and given trademark protection is typi-
`cally shorter and easier to remember than the generic name.
`The pharmaceutical industry promotes its products heavily. For many
`research-based firms the promotion budget can be twice to four times as
`large as the budget for research and development, with sales promotion
`running 20 to 30 percent of sales.8 The most heavily used form of promo-
`tion consists of visits to physicians, pharmacists, and other health-care
`professionals by sales representatives of the producers of branded phar—
`maceuticals. Almost 70 percent of the pharmaceutical industry‘s promo—
`tional budget is spent on personal promotion, known as “detailing.” An-
`other 2? percent is spent on advertising in the many journals addressed to
`physicians, who receive on average between seven and twentyjournals a
`month.9 Direct mail accounts for the balance of measured promotional
`outlays. A number of activities of pharmaceuticals firms are geared less
`toward promoting a specific drug than enhancing the general reputation of
`the company. These include textbooks, audiovisual aids, and lectures and
`seminars sponsored as part ofcontinuing medical education for physicians.
`There is evidence that the goodwill asset built up for a patented drug
`substantially outlasts the patent protection. Bond and Lean examined the
`development of two therapeutic markets, orally effective diuretics and
`antianginals. In each, the first firm to offer a new product was able to
`maintain a substantial market share despite entry into the therapeutic
`class of other drugs that were priced lower and promoted more heavily,
`although late entrants tended to be more successful when their brands
`offered some therapeutic novelty.IO
`After the patent’s expiration, sales-promotion outlays and other differ-
`entiation strategies remain open to branded producers. Certain qualitative
`strategies of product differentiation are especially useful in the postpatent
`period. The delivery system—the system required to deliver medication
`to the part of the body where a therapeutic effect is desired—is a common
`element of product differentiation. Differentiated oral dosage forms in-
`
`” Drug Product Selection. Staff Report to the Federal Trade Commission, Bureau of
`Consumer Protection 32 (1979]; Comanor, supra note 1, at [196.
`9 Drug Product Selection, supra note 8, at 58.
`‘0 Ronald S. Bond & David F. Lean, Sales, Promotion, and Product Differentiation in
`Two Prescription Drug Markets, Staff Report, Bureau of Economics, U.S. Federal Trade
`Commission ([977). The value of these goodwill aSsets is also confirmed by the evidence that
`most (excess) profits earned by the pharmaceutical industry are due to new drugs. See
`Martin N. Baily, Research and Development Costs and Returns: The U.S. Pharmaceutical
`Industry, 80 J. Pol. Econ. T0 ([972].
`
`IMMUNOGEN 2288, pg. 4
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROMOTION OF PHARMACEUTICALS
`
`303
`
`clude time-release capsules and enteric coated pills. Taste can be an
`important factor, particularly tO ensure patient compliance. Differentia—
`tion in packaging, such as the provision of prepackaged unit doses and
`special packaging to fit hospitals‘ dispensing units, provides an additional
`opportunity for protecting the market for patent-lapsed products. The
`appearance of a trademarked pill apparently can be protected from imita-
`tion by an entrant; furthermore, the branded manufacturer can form its
`pill in a shape that is linked to the registered trade name.
`
`Producers of Generic Drugs
`
`The new producer of a drug is foreclosed from using the discoverer’s
`trademarked name but can choose between promoting its own brand
`name (Often an amalgam Of the generic name and the name of the prO~
`ducer] or simply offering the drug under its generic name. While some of
`the larger producers of “commodity“ generics promote the names of their
`companies, most do not.
`Drug innovators face price-inelastic demands and accordingly set
`prices markedly in excess of marginal costs. ” Entrants therefore have the
`opportunity to quote prices that exceed their marginal costs while offering
`large discounts from the patent holder‘s price. Discounts average at least
`20 percent and may leave generics’ prices to final consumers as much as
`two-thirds below the original branded drug’s.‘2 While small retail phar—
`macies usually pay prices close to the published list, prices paid by phar—
`macy chains or buying groups, health maintenance organizations, hospi-
`tals, and other large buyers, such as governments, can vary widely. Both
`brand and generic manufacturers sell their products to hospitals at dis-
`counted prices that can fall as low as 25 percent of prices to pharmacies.
`Bid and group purchasing by hospitals can dramatically lower the prices
`paid for pharmaceuticals.I3 Thus, price differentials selected by entrant
`generic producers and responses by the developers of branded drugs
`
`11 The evidence confirms our expectation that new drug products with any therapeutic
`novelty set prices that are high relative to their established competitors in the therapeutic
`class, but these relative prices drop over the next few years. See W. Duncan Reekie, Price
`and Quality Competition in the United States Drug Industry, 26 J. Indus. Econ. 223 (1973}-
`” In our own sample (described below) the average discount for generics at wholesale is
`56 percent. Prices vary among entrants, with branded generics and products of the larger
`manufacturers carrying premia over the smaller generic companies. A [980 sample of thirty—
`seven multisource drugs found retail prescription prices for generics to average 24 percent
`lower. See Alison Masson 8: Robert L. Steiner, Generic Substitution and Prescription Drug
`Prices: Economics Effects of State Drug Product Selection Laws 36 (1985).
`[3 A [978 study showed that nineteen multisource products were on average 31 percent
`less expensive when purchased by bid. See Frost & Sullivan, Generic Drugs, [OS—4(191r'9).
`
`IMMUNOGEN 2288, pg. 5
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`304
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`represent one form of strategic interaction once a patent lapses. The
`available evidence suggests that sequences of pricing moves associated
`with expiring patents are diverse, with branded producers making antici-
`patory price cuts in some cases and declining to match entrants‘ prices in
`others.”
`
`Whether drugs made by branded and generic producers are equivalent
`in quality is an important and controversial issue. The method used by
`medical researchers to compare the quality of chemically identical drugs
`is that of measuring their “bioavailability,” or how much of the drug‘s
`active ingredient gets into the bloodstream and to the site of therapeutic
`action, and at what rate. Two chemically identical drugs that have ap-
`proximately equal bioavailability are said to be bioequivalent. If they are
`equally effective in treating a particular disease, they are said to be thera-
`peutically equivalent. The U.S. Food and Drug Administration (FDA), in
`its bioavailabilityi’bioequivalence regulations of 1977, established criteria
`to identify products subject to bioequivalence problems and developed
`procedures by which manufacturers of such earmarked products must
`demonstrate their equivalence. Authorities in the field of biopharmaceu-
`tics, contacted by the U. S. Federal Trade Commission (FTC}, believed
`that the criteria used by the FDA to develop its list of problem drugs
`represent the best possible scientific judgment currently available. The
`FDA itself concluded that “there is no substantial evidence of significant
`differences, either in bioavailability or general quality (in terms of purity,
`potency, or other methods of quality control) between brand-name and
`unbranded products, or between products made by large and small manu—
`facturers.” '5 Although the FDA’s list of approved drug products provides
`information regarding the therapeutic equivalence of virtually all prescrip—
`tion drugs, it is not the only source of this information. Preseribers can
`draw on many other nonpromotional sources ofinformation on the quality
`of competing products.
`A separate issue from average product quality is that of quality control
`
`1‘ David Schwartzman, Innovation in the Pharmaceutical Industry 296—98 (1976}, Meir
`Statman, The Effect of Patent Expiration on the Market Position of Drugs, in Drugs and
`Health: Economic [ssues and Policy Objectives 140(Robert B. Helms ed. [98]). A Canadian
`study suggests that branded drugs‘ prices are typically unresponsive, but that may reflect
`special circumstances; see James J. McRae & Francis Tapon, Some Empirical Evidence on
`Post—patent Barriers to Entry in the Canadian Pharmaceutical Industry, 4 J. Health Econ. 43
`(1985). Theoretically, the wealth-maximizing price differential maintained by the original,
`branded producer could either increase or decrease over time following entry, with increase
`possible if the more price—sensitive buyers are the first to switch to cheaper generics, so that
`the elasticity of substitution fails as generics’ share increases. The more likely sounding case
`of a declining margin between generic and branded drugs would result if the elasticity of
`substitution increases over time or if the time-rate of entry of new producers is constrained.
`‘5 Drug Product Selection, supra note 8, at 245, 256.
`
`IMMUNOGEN 2288, pg. 6
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROMOTION OF PHARMACEUTICALS
`
`305
`
`and manufacturing practice. Schwartzman urged that there are theoretical
`reasons why smaller generic producers might invest less in quality control
`than larger producers of numerous branded drugs.” The FDA finds pro—
`portionally more problems with recalls and failed inspections among the
`smallest producers; however. the evidence from laboratory tests on final
`products does not clearly weigh against generics.” There is no evidence
`that generics‘ clinical performance is at all affected by differences in
`quality control.
`
`Health Professionals
`
`The key choices of drugs are typically made by physicians, who pre-
`scribe drugs to their patients and leave the actual dispensing to retail or
`hospital pharmacists. When writing a prescription for a multisource drug,
`the physician can prescribe it by either the brand or the generic name. The
`physician has no substantial economic incentive to choose the IOWer—
`priced product, and doctors tend to be ignorant of specific drug prices.
`The brand name, simpler than the generic name, was learned when the
`drug was introduced and is easier to remember, and habit accordingly
`plays a strong role in the physician‘s prescription practice.is
`There is also the question of physicians‘ perceptions of the qualities of
`branded and generic drugs. Surveys indicate that physicians place less
`faith in the quality control of generic than of branded producers; almost
`half of those surveyed disagreed with a statement that small generic
`manufacturers maintain quality—control standards equal to those of brand
`name companies. Escalating malpractice suits have prompted risk-averse
`physicians to prescribe higher—priced brands that are trusted on the basis
`of experience or reputation rather than risking an ineffective treatment or
`adverse reaction from a poor-quality drug. Accordingly, it is no surprise
`that the rate of generic prescribing is low, with 11.6 percent of all pre-
`scriptions filled by retail pharmacists in 1978, rising only to 14.7 percent in
`1983.“31
`
`‘6 Schwartzman, supra note [4, at 215—23. Quality control is likely to entail a substantial
`element of fixed cost, and the smaller or specialized producer has less to lose from negative
`goodwill.
`” Schwartzman, supra note 14, at 226—46.
`18 Drug Product Selection, supra note 8, at 63, I54. Accordingly, Masson & Steiner. supra
`note II, at 89, 101, found the form of the physician‘s prescription pad the strongest
`identifiable determinant of whether generic substitution is permitted in a prescription. Also
`see Henry G. Grabowski & John M. Vernon, Substitution Laws and Innovation in the
`Pharmaceutical Industry, 43 Law & Contemp. Probs. 43 (19%)}.
`‘9 Pharmacy Times Annual Survey, based on data from [MS America. Most brand name
`drug producers (and only some generic producers) offer to pay for damages incurred as a
`result of properly prescribing their products.
`
`IMMUNOGEN 2288, pg. 7
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`306
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`The retail pharmacist’s role in the selection of drug products has been
`increasing as physicians write more generic prescriptions and as states
`have repealed antisubstitution laws that used to make it illegal to substi-
`tute a generic copy for a prescription that specifies a brand name.10 Phar-
`macists' training prepares them to make choices among chemically identi—
`cal products, but their incentives for minimizing cost to the consumer also
`are not necessarily strong. To the extent that pharmacies1 retail prices
`involve fixed percentage markups, their incentives run contrary to filling a
`prescription at the lowest cost. In 1979, laws in nineteen states required
`that the whole saving from utilizing a generic drug be passed along to the
`customer.ll Perfect competition among pharmacies in supplying well-
`informed buyers would of course insure resorting to generics whenever
`the prescription permits and the buyer wishes; however, shopping costs
`for customers with other than chronic needs surely limit this competitive
`force.22 Perhaps as a result, pharmacists‘ substitution of generic for
`branded drugs, where this is permitted by the prescription, has risen only
`to approximately 16 percent of eligible prescriptions.23
`When drugs are prescribed and dispensed in hospitals, the incentives
`and informational capabilities Ofthe various actors differ from those in the
`case of drugs sold by retail pharmacies. The hospital formulary system is
`based on a contract between the physician and the hospital by which the
`physician agrees that prescriptions written under a brand name shall be
`interpreted as if they were written generically, unless the physician indi—
`cates otherwise. The hospital pharmacist is encouraged to prescribe those
`products listed on the hospital’s formulary. The formulary is a continu-
`ously revised list of drug products approved by the hospital‘s pharmacy
`and a therapeutic committee consisting of pharmacy, clinical, and nursing
`staff members.“ This system pools information on the cost and effec-
`
`1a Physicians may indicate on the prescription form that no substitution may be made for
`a branded drug; the degree to which the form facilitates this (and also the frequency with
`which it is done] varies from state to state. Drug Product Selection, supra note 8, at [ST—90.
`11 Drug Product Selection, supra note 8, at 13‘7—82.
`22 As of19T9,twenty‘twO states required the pharmacist to notify the patientofasubstitu-
`tion, and seventeen recognized the patient‘s right to veto it. Drug Product Selection, supra
`note 8, at [68. Surveys ofconsurners‘ attitudes toward generics give inconsistent results but
`suggest that they are not well informed and may associate higher price with higher quality.
`23 Masmn & Steiner, supra note 20, at 30—32, projected this figure for 1984. They showed
`(at 37—40) why pharmacists‘ markups should be higher on generic than branded prescrip—
`tions, as their average data confirm. See also J. P. Gagnon & C. A. Rodowskas, Reimburse-
`ment Methods for Pharmaceutical Service, [4 .I. Am. Pharmaceutical Soc. 675 (l9'i'4).
`2“ In 1982 the formulary system was in effect in 90.3r percent of U.S. hospitals. Michael H.
`Stolar, National Survey of Hospital Pharmaceutical Services—4982, 40 Am. J. Hospital
`Pharmacy 966 (I983). Leffler, supra note 4, at 53—54, noted the unimportance of sales
`promotion in the hospital markets.
`
`IMMUNOGEN 2288, pg. 8
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROMOTION CIF PHARMACEUTICALS
`
`307
`
`tiveness of different drugs and helps hospitals to minimize drug costs.
`Furthermore, hospitals and other inpatient facilities have come under
`substantial pressure to minimize costs. Public and third-party payers with
`an interest in cost minimization cover 8?. percent of hospitals' drug expen-
`ditures.25
`
`The federal government, the largest third-party purchaser of pharma-
`ceuticals, employs the Maximum Allowable Cost (MAC) program, which
`limits reimbursements for a multisource drug under federally funded pro-
`grams, such as Medicaid, to the lowest price at which a generically mar-
`keted drug is widely available to pharmacists. Many states have programs
`of their own similar to MAC. These programs significantly increase ge—
`neric substitution by pharmacists.26
`
`Conclusions
`
`While this evidence identifies some factors that have shifted to enhance
`
`the market access of generic entrants, overall it confirms the importance
`of information costs and constraints for competition in the pharmaceutical
`market. We therefore proceed to the test, outlined in this article’s in—
`troduction, of the interdependence of incumbents‘ postpatent market
`shares, sales-promotion outlays, and pricing decisions and their joint de-
`pendence on the drug’s customer base and the type of therapy for which it
`is used. Because the evidence reviewed in this section suggests that
`branded-drug shares should be more difficult to preserve in the hospital
`market than in the retail pharmacy market, we also test whether branded-
`drug shares are less sensitive to their advertising outlays in the hospital
`market.
`
`11.
`
`STATISTICAL MODEL
`
`We constructed a sample by drawing from a list of 150 drug products
`that are available as generics but were originally subject to patents held by
`the firms that developed them.27 Products not available in oral dosage
`forms were eliminated in order to observe products that are widely avail—
`able in both hospitals and pharmacies. A sample of twenty-nine separate
`
`93 W. M. Allen, Prospective—Payment Systems: The Process Becomes Reality, 4 Pharma—
`ceutical Executive 56 (May [984).
`
`’5 Masson & Steiner, supra note [2, at 53—54, 104. The effect of similar programs in
`Canada on branded producers‘ abilities to sustain both high share and high prices was shown
`by McRae & Tapon. Supra note 14.
`’7 In some cases branded products were initially marketed simultaneously by two firms
`due to cross licensing or dual marketing arrangements. We assumed that these partners
`could be regarded as joint monopolists during the patent period; data for the two were
`
`IMMUNOGEN 2288, pg. 9
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`308
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`multisource drug markets was obtained, seven of which are among the top
`twenty-five generically prescribed drugs. Because we are interested in
`observing drug markets at varying numbers of years after their patents
`have expired, as many as possible of these drug markets were observed in
`both 1978 and 1983, providing a sample of fifty-six observations.
`
`Variables to Be Explained
`
`Two variables are to be explained in the following analysis: the market
`share held by the original patent holder (hereafter the “leading firm") and
`the ratio of that firm’s sales promotion of the drug to the value of its sales.
`In order to calculate the former, we had to address the problem of the
`diverse strengths and dosage forms in which drugs are typically marketed.
`We chose to measure the market share in physical units—number of pills
`sold by the leading firm divided by total pills sold. Because of large price
`differentials between the leader’s and others“ products, a physical mea-
`sure is clearly preferable to a share of dollar sales. However, it is also not
`appropriate to combine doses marketed in different forms and strengths,
`because (for one reason] the FDA‘s information on equivalency pertains
`only to pills of identical dosage form and strength. When a preliminary
`analysis of the data indicated that market shares of different dosage forms
`of the same product do not differ significantle the decision was made not
`to consolidate sales data for different dosage forms and strengths. In-
`stead, we selected the most popular dosage form (the one that had the
`highest dollar sales for the greatest number of producers) and con
`solidated only different package sizes for that dosage before determining
`the leader‘s market share.28 The variable is designated SHARE.
`The other variable to be explained is promotion of the drug by the
`leading firm in the year in question. A firm‘s sales—promotion outlay was
`calculated as dollars spent on detailing, medical-journal advertising, and
`direct-mail advertising. We secured this outlay for each drug product for
`both the leading firm and (as an exogenous variable—see below) the
`aggregate of other sellers.” The leader‘s outlay was divided by that firm‘s
`
`pooled, and they were treated jointly as the leading brand. The list was taken from Frost &
`Sullivan, supra note 13, at 48. It is worth noting that many off-patent drugs are not chal-
`lenged by entrants—thirty—four of fifty-two in the sample of Henry Grabowski 8t John
`Vernon, Longer Patents for Lower Imitation Barriers: The 1984 DrUg Act, ”l6 Am. Econ.
`Rev. Papers & Proc. 195 (1986).
`2” Sales data were obtained from the U.S. Pharmaceutical Market Drugstore and Hospital
`Audits, published by [MS America, Ltd. [MS estimates sales through a nationwide audit of
`retail drugstores and hospitals, the results of which are validated by comparison with factory
`shipments figures.
`19 Data were obtained from the National Detailing, Journal, and Mail Audits published by
`[MS America, Ltd. [MS estimates detailing expenditures by surveying a sample of physi-
`
`IMMUNOGEN 2288, pg. 10
`Phigenix v. Immunogen
`|PR2014—00676
`
`IMMUNOGEN 2288, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROMOTION OF PHARMACEUTICALS
`
`309
`
`total sales of the drug (in all dosage forms) to obtain the variable AD-
`SALES.
`
`Exogenous Variables Aflecting Share
`
`Our Specification of models to explain SHARE and ADSALES fOIIOWs
`the same general strategy as other studies ofthis type30 but with variables
`selected and specified to reflect the particular features of drug markets.
`Our model of SHARE is based on the assumption that demand for each
`type ofa drug—leader‘s brand or generic substitute—decreases with own
`price and rivals” promotional expenditures and increases with rivals‘ price
`and own promotion. When those demand functions are consolidated to
`explain SHARE, we expect it to depend on the relative price, expressed
`as the prOportional price discount offered by competing producers
`(PRDIF). To calculate PRDIF we obtained the weighted average invoice
`price paid by drugstores to their suppliers for the leading firm’s pills (PL),
`the median of the average prices paid by drugstores for generic pills (PF).
`and calculated PRDIF 2 (PL — PFUPF.31 Sales—promotion outlays by the
`leader (LPROM) and by all competing suppliers (FPROM) enter the
`model, however‘ not as a ratio but as logarithms of the respective dollar
`amounts. Because the leader‘s and rivals‘ sales—promotion dollars may
`not be equally effective in their influence on market share, we allow them
`to take different coefficients. Furthermore,
`this specification as loga—
`rithms of absolute amounts (not normalized) is appropriate for a cross—
`section sample of pharmaceuticals because all drugs in the sample are
`promoted to essentially the same market of prescribers.32
`Several variables capture the extent of brand loyalty built up by the
`
`cians and collecting data on visits made by sales representatives and minutes spent per drug.
`The figures are transformed to dollars by spreading an estimate of cost per call over the
`minutes spent for each drug. Thejournal audit analyzes 350 publications for advertisements
`and uses published rates to determine expenditures. The mail audit analyzes mailings re,
`ceived by a sa