THE JOURNAL OF INDUSTRIAL ECONOMICS
`
`Volume LII
`
`September 2004
`
`0022-1821
`
`No. 3
`
`WHAT EXPLAINS THE USE OF DIRECT-TO-CONSUMER
`ADVERTISING OF PRESCRIPTION DRUGS?*
`
`TOSHIAKI IIZUKA'l‘
`
`Following the clarification of advertising regulation in 1997, direct—to—
`consumer advertising (DTCA) of prescription drugs has skyrocketed in
`the US, creating a controversy over the role of DTCA. Little is known,
`however, regarding what affects firms’ advertising decisions and which
`drugs have been advertised to consumers. Using brand—level advertising
`data, I examine the determinants of DTCA of prescription drugs. I find
`that drugs that are new, of high quality, and for under—treated diseases
`are more frequently advertised. Furthermore, advertising outlays
`decrease with competition. These results complement the demand—side
`evidence that DTCA has a market—expanding effect but little business—
`stealing effect.
`
`I.
`
`INTRODUCTION
`
`Direct-to-consumer advertising (DTCA) of prescription drugs had been
`viewed as taboo for a long time. Traditionally, pharmaceutical firms have
`promoted their prescription drugs through detailingithe face-to-face
`selling by medical representatives directly to physicians. The clarification of
`advertising regulations in 1997, however, changed this tradition drastically.
`Now,
`firms can use product-specific television commercialsiwhich
`mention both the name and the use of the drugito promote their
`prescription drugs to consumers without fully disclosing the risks of the
`drugs. As a result, within only five years after the clarification of regulations,
`prescription drug advertising expenditures skyrocketed from $800 million in
`1996 to $2.7 billion in 2001.
`The dramatic increase of DTCA has created a new controversy over the
`role of advertising in the prescription drug market. The main argument in
`favor of the new policy is that consumers can gain valuable information
`through DTCA. It is argued, for example, that advertising can inform
`patients about new medications for diseases that were believed to be
`
`*I wish to thank Dan Ackerberg, Bill Comanor, Austun Goolsbee, V. Joseph Hotz, Tom
`Hubbard, Phillip Leslie, Jean—Laurent Rosenthal, the editor and an anonymous referee for
`helpful comments and suggestions. I am grateful for TNS Media Intelligence/Competitive
`Media Reporting (CMR) for generously providing the data for this study. Remaining errors
`are my own.
`
`TAuthor’s affiliations: Owen Graduate School of Management, Vanderbilt University, 401
`21SI Ave. South, Nashville, TN 37203, USA.
`e—mail: toshi.iizuka@owen. vanderbilt.edu
`© Blackwell Publishing Ltd. 2004, 9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden, MA 02148, USA.
`349
`
`Argentum Pharm. LLC v. Alcon Research, Ltd.
`Case IPR2017-01053
`
`ALCON 2070
`
`

`

`350
`
`TOSHIAKI IIZUKA
`
`untreatable by medicines. On the other hand, critics are concerned, for
`example, that DTCA may affect the choice of treatments by providing
`information of suspect quality and encourage people to try more expensive
`drugs though equally effective, but cheaper, drugs may be available.
`Responding to these concerns, the Food and Drug Administration (FDA)
`has recently announced that it will review its policy on DTCA.1
`Despite the surge of DTCA, its potential effects on consumer health, and
`the intensive policy debates, economics research on DTCA of prescription
`drugs is scarce. In particular,
`little is known about what affects
`pharmaceutical firms’ advertising decisions and which drugs have been
`advertised to consumers. The main objective of this paper is to fill this gap by
`analyzing the determinants of DTCA. A striking feature of DTCA is that,
`unlike detailing promotion, it is concentrated on a small number of drugs in
`some specific therapeutic categories. Using a unique panel data set that
`contains more than 600 drug-year observations over 1996–1999, I examine
`when and why firms advertise. To this end, a censored regression model,
`which takes into account zero advertising expenditure by many firms, and a
`two-stage model, which allows for a qualitative difference between ‘whether
`to advertise’ and ‘how much to advertise’ are estimated. I make reference to
`various classes of advertising theories to guide the empirical analysis.
`To be sure, the main reason for the lack of research is that DTCA of
`prescription drugs is only a recent phenomenon. On the demand side,
`however, a few recent papers have started exploring the effects of DTCA.
`Rosenthal et al. [2003] examine the effects of DTCA on the sales of six
`therapeutic classes and find that DTCA has a significant effect on aggregate
`demand but does not affect market shares within each class. Similarly,
`Iizuka and Jin [2003] find that DTCA leads to a large increase in outpatient
`visits, but has no effect on doctors’ specific choices among prescription drugs
`within a therapeutic class. Wosinska [2002] focuses on cholesterol-reducing
`drugs and finds that DTCA affects the demand for an individual brand
`positively, but the impact is substantially smaller than that of detailing
`promotion. All of these papers suggest that DTCA may have a large market-
`expanding effect but little or no business-stealing effect.
`On the supply side, as noted before, little research exists on DTCA of
`prescription drugs. Previous papers have examined, instead, various aspects
`of detailing promotion. Leffler [1981] observed a cross-section of 35
`therapeutic categories (not individual drugs) and examined the differences in
`detailing intensity across the categories. He found that empirical results are
`consistent with both ‘informative’ and ‘persuasive’ views of advertising.
`Hurwitz and Caves [1988] looked at a cross-section of 56 off-patent drugs
`
`1 The Wall Street Journal, ‘FDA to Review Policy Allowing Drug Ads on TV,’ March 28,
`2001.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING OF PRESCRIPTION DRUGS
`
`351
`
`and analyzed the determinants of detailing intensity. They found that,
`among other findings, branded products’ detailing intensity decreases as the
`number of generic competitors increases.
`Several empirical results are worth noting. First, I find that firms are more
`likely to advertise newer and higher-quality drugs rather than older and
`lower quality ones, other things being equal. The latter indicates that DTCA
`and product quality complement each other in this market. Second, firms
`advertise more when the number of potential patients, rather than currently
`treated patients, is large. This complements the demand-side evidence that
`DTCA has a market-expanding effect but little or no business-stealing
`effect. This result is also consistent with the claim of proponents that DTCA
`targets under-diagnosed therapeutic classes and, thus, could be welfare
`improving. Third, I find that firms advertise less when therapeutic and
`generic competition gets intense. This suggests that DTCA does not have a
`strong effect to shift market shares among alternative drugs, which is also
`consistent with the demand-side finding discussed above. Lastly, I find early
`entrants are more likely to advertise than late entrants. This suggests that the
`return from DTCA is higher for early entrants, i.e., ‘first mover advantages’
`in DTCA appear to exist in this market.
`The remainder of the paper is organized as follows: Section II briefly
`reviews regulations and controversies on DTCA. Section III discusses the
`potential determinants of DTCA. After describing the data and variables in
`the next section, Section V discusses estimation and identification issues.
`Section VI presents results, and Section VII discusses alternative explana-
`tions. Section VIII concludes the paper.
`
`II. PRESCRIPTION DRUG ADVERTISING: REGULATION AND CONTROVERSY
`
`II(i). An Overview of Advertising Regulation
`
`Promoting prescription drugs directly to consumers is a recent phenomen-
`on. Traditionally, prescription drugs have been marketed exclusively to
`physicians either by detailing promotion, or, to a lesser extent, by advertising
`in medical journals. Pharmaceutical firms assumed that doctors would never
`accept a program that bypassed them, and DTCA was conceived as suicidal
`(Pines [1999]).
`In the early 1980s, however, a few firms started advertising their products
`directly to consumers. The FDA took this seriously and asked the industry
`for a voluntary prohibition period during which the FDA would study the
`impact of DTCA on public health. In 1985, the FDA announced that current
`regulations, the Kefauver-Harris drug amendments of 1962, were sufficient
`to protect consumers. This meant that, as long as manufacturers provided a
`‘brief summary’ of contraindications, side effects, and effectiveness and
`maintained ‘fair balance’ among them, DTCA would be permissible. The
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`352
`
`TOSHIAKI IIZUKA
`
`FDA appeared to move in this new direction because, for one reason, it
`recognized that consumers increasingly wanted to obtain more information
`about prescription drugs (Pines [1999]).
`Not surprisingly, DTCA increased thereafter but was mostly limited to
`newspapers and magazines because of the ‘brief summary’ requirement.
`Providing the ‘brief summary’ is costly for firms since it commonly occupies
`a full page or more of magazine space even though firms use very tiny fonts to
`describe them. The FDA essentially required TV advertising to abide by the
`same rule, and thus DTCA was prohibitively expensive for TV media.
`Accordingly, firms did not often use TV commercials to promote
`prescription drugs.
`There were two conditions, however, under which firms could avoid
`the ‘brief summary’ in TV advertising. One was the so-called ‘help-seeking’
`ad in which only disease symptoms were mentioned but not the specific name
`of the drug. The other was when the firm mentioned only the name of the
`drug without saying what it was for. The use of these types of ads
`continuously increased during the mid 1990s. The rapid growth of
`Health Maintenance Organizations (HMOs) and the increase of break-
`through drugs might have encouraged firms to use this new channel of
`communication.
`It was not until 1997, however, that a breakthrough occurred when the
`FDA further relaxed its regulation of ethical drug advertising on TV. For the
`first time, the FDA permitted product-specific DTCA on TV, which
`mentioned both the drug’s name and the condition for which it was to be
`used, without disclosing the ‘brief summary.’ Now firms needed only to
`include ‘major statements’ of the risks and benefits of the drug, which
`required substantially less information and airtime. Thus, by reducing the
`cost of advertising, the policy change contributed to the surge of DTCA after
`1997. Pines [1999] explains that the FDA made this change because it
`recognized that ads that mentioned a drug’s name but not its use were non-
`communicative and even confusing to consumers. Wilikes et al. [2000] also
`point out that ‘the political and regulatory climate was moving toward
`allowing consumers more choice and empowering them to share in medical
`decision making.’
`An interesting feature of DTCA is that the FDA assumes jurisdiction over
`it because the FDA views DTCA as a ‘label,’ a package insert describing the
`characteristics of the drug. Accordingly, the FDA monitors and enforces
`information contents of DTCA quite vigorously.2 In fact, pharmaceutical
`firms often ask the FDA to review their advertising commercials before they
`launch an advertising campaign. Because of these interactions, as well as the
`
`2 The FDA has threatened violating firms with legal actions,
`including seizure and
`injunction. Pines [1999] discusses the history of the FDA’s enforcement activities in detail.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING OF PRESCRIPTION DRUGS
`
`353
`
`‘major statements’ and ‘fair balance’ requirements, prescription drug
`advertising is likely to convey credible information on drug attributes.
`
`II(ii). The Effects of the Relaxation of Advertising Regulation
`
`Following the FDA clarification in 1997, DTCA of ethical drugs increased
`dramatically. Within five years of the clarification, DTCA surged from $800
`million in 1996 to $2.7 billion in 2001. The surge of DTCA, however, was not
`observed equally across drugs and therapeutic classes. On the contrary,
`firms have been very selective in the use of DTCA. In 1999, approximately
`41% of total DTCA ($1.8 billion) was spent for the top ten advertised drugs,
`while their sales share was only 9%.3 Why do firms sometimes use DTCA to
`promote their drugs but not always? I will discuss some potential
`determinants of DTCA in this market in the next section.
`Anecdotal evidence shows that DTCA has indeed encouraged potential
`patients to seek medical help. Based on a national survey conducted in 1998,
`Prevention magazine found that DTCA encouraged a projected 21.2 million
`consumers to talk with their doctors about a medical condition or illness
`they had not previously talked with their doctor about before seeing an
`advertisement. Furthermore, the magazine estimates that 12.1 million
`people received a prescribed drug as a direct result of seeing a DTC
`advertisement. A Time survey conducted in 1998 also shows that one-fourth
`of consumers who saw an advertisement on television or in a magazine and
`spoke with their physicians about it received a prescription.
`
`II(iii). Controversies
`
`The tremendous increase in DTCA and prescriptions in recent years has
`created a major controversy over the effects of such advertising on
`pharmaceutical demand. In particular, two distinct views exist on the
`effects of DTCA. Proponents of DTCA argue that the match between
`patient and drug could be improved if consumers were informed about
`prescription drugs through direct consumer advertising (Masson and Rubin
`[1985]). They also argue that direct advertising plays an important role in
`informing the public of the existence of treatments of diseases, some
`previously not believed to be treatable by medicines (Masson and Rubin
`[1985]; Holmer [1999]). It is known that a number of leading diseases,
`including diabetes, high-cholesterol, and high-blood pressure, are under-
`diagnosed or under-treated. Thus, they argue, DTCA could help improve
`the health of people with these conditions. Holmer [1999]
`further
`
`3 DTCA figures are from IMS Health’s press release,
`‘IMS Health Reports U.S.
`Pharmaceutical Promotion Spending Reached Record $13.9 billion in 1999,’ on April 20,
`2000. Sales figures are also from IMS Health reported in Pharmacy Times, ‘The Top 200 Drugs
`of 1999,’ http://www.pharmacytimes.com/ top200.html.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`354
`
`TOSHIAKI IIZUKA
`
`hypothesizes that ‘DTCA merely motivates patients to learn more about
`medical conditions and treatment options and to consult their physicians,
`but once the dialogue is started, the physician’s role is preeminent (page
`381).’ In sum, proponents claim that DTCA provides valuable information
`to consumers and increases physician visits. They argue that DTCA does
`not, however, affect the choice of prescription.
`On the other hand, opponents of DTCA are concerned that it may affect
`the choice of prescription and increase the cost of services. Hollon [1999]
`argues that DTCA may provide information of suspect quality and, ‘by
`creating consumer demand, undermine the protection that is a result of
`requiring a physician to certify a patient’s need for a prescription drug.’
`Others also argue that DTCA may encourage people to try more expensive
`drugs although equally effective, but cheaper, drugs may be available
`(Cohen [1988]). Further, it is often reported that many physicians fear that
`under-informed patients will demand inappropriate therapies from doctors
`once they have seen DTCA. That is, opponents argue that DTCA may
`manipulate the choice of prescription, and this could be harmful for patients
`and increase medical costs as well. Clearly, the source of the controversy is
`the role of DTCA in this market.
`While the main objective of this paper is to examine the determinants of
`DTCA and not to distinguish between these two claims, nonetheless, some
`of the results of the paper might help clarify the validity of these claims. For
`example, I consider whether DTCA outlays are more responsive to ‘current’
`market size as opposed to ‘potential’ market size. Also, I examine whether
`competition with generic drugs would increase brand-name drugs’
`advertising outlays. If the answers are ‘yes’ to these questions, the results
`are less likely to support the proponents’ claim since the ‘informative’ role of
`advertising is secondary in this situation. Now, I turn to the analytical
`framework to examine the determinants of DTCA.
`
`III. DETERMINANTS OF DTCA OF PRESCRIPTION DRUGS
`
`DTCA in the prescription drug market is unique in that the use of
`advertising is concentrated on a small number of products. This section
`discusses various classes of advertising theories that may help explain why
`and when firms use DTCA to promote their prescription drugs.
`
`III(i). Competition and DTCA
`
`Brand Competition: Economists have long debated the effects of competi-
`tion on advertising (and vice versa).4 Earlier literature on this issue often
`
`4 A large number of earlier papers examined inter-industry relationship between profitability
`(or concentration) and advertising intensity (e.g. Telser [1964]; Schmalensee, [1972]; Comanor
`and Wilson, [1974]).
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING OF PRESCRIPTION DRUGS
`
`355
`
`found that advertising decreases as competition gets intense. For example, in
`their classic paper, Dorfman and Steiner [1954] showed that a monopolist’s
`advertising intensity decreases as demand elasticity increases, other things
`being equal. Based on this result, it is often argued that advertising intensity
`should decrease with competition, since demand is more elastic in comp-
`etitive markets.5 More recently, Grossman and Shapiro [1984] examined the
`effects of competition on advertising in the case of differentiated product
`oligopoly. They show that in the case of ‘informative’ advertising – which
`informs consumers about the existence and characteristics of the product –
`firms reduce their advertising when there are more close substitutes in the
`market. This happens because, as the number of close substitute increases,
`consumers are likely to receive an ad from a firm located close to the
`consumer, whose product provides a better match between the product and
`patient. This in turn reduces firms’ incentive to advertise.6 Moreover, if
`DTCA has a market-expanding effect but not a business-stealing effect as
`the demand-side papers suggest, then advertising should decrease with
`competition since competitors may be able to free ride on rival firms’
`advertising.
`Others argue, however, that firms increase advertising when competition
`becomes intense. Becker and Murphy [1993] argue that, as close substitutes
`increase, firms may try to differentiate themselves by using advertising, and
`this will lead to higher advertising expenditures. For example, they discuss
`that ‘Perdue chicken’ (and other products in a competitive market) is
`extensively advertised because ‘Perdue ads convince consumers that a pound
`of its chicken is worth more than a pound of other chickens.’ (pp. 954–55).
`Cabral [2000] also notes that as the number of competitors increases, each
`firm’s incentive to engage in business-stealing advertising may increase. This
`may be the case because the return from such advertising may increase as the
`residual demand increases. In summary, DTCA of prescription drugs could
`either increase or decrease as competition gets intense: which effect
`dominates the other is an empirical question.
`
`5 Becker and Murphy [1993], however, revisit the Dorfman and Steiner model and show that
`the earlier discussion was sensitive to the assumption, and incentive to advertise may indeed
`either increase or decrease with competition.
`6 Grossman and Shapiro [1984] assume that consumer has no alternative sources of
`information, and is unaware of the existence of a particular brand unless she sees an
`advertisement describing it. In addition, consumers are assumed to remember all advertising
`messages transmitted to them. Under these assumptions, they show informative advertising
`decreases with competition. I believe these assumptions are not unreasonable in the case of
`prescription drugs, especially for newly discovered drugs, because it is costly for consumers to
`obtain information on prescription drugs. However, if we relax these assumptions, above-
`mentioned results may not hold. For example, if patients do not know or remember the
`characteristics of the drug, a monopolist may wish to (re)inform the patients about the drug’s
`characteristics once a rival drug enters the market. In this case, informative advertising may
`increase with competition.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`356
`
`TOSHIAKI IIZUKA
`
`Generic Entry: It is well known that generic entry has a major impact on the
`market share of brand-name drugs. Typically, within a few months after
`patent expiration, generic entry takes place, and firms sell their generic drugs
`as low as 80% below the branded drug’s former list price (S&P [1999]). While
`the price response to generic entry of branded products has been a source of
`controversy,7 market share of branded drugs typically drops substantially
`due to drastic demand shift to generics.8 As a result, revenue of branded
`drugs often drops sharply after patent expiration.
`To protect their profits, incumbents may increase DTCA to differentiate
`the brand-name drug from generics. Again, the motivation is very similar to
`the ‘Perdue chicken’ example discussed above and thus requires no further
`explanation. Note, however, that advertising is likely to be wasteful in this
`case since generics are mostly identical to the branded drug in terms of
`pharmacological benefits to patients. Unlike the case of competition among
`brands discussed above, advertising cannot improve the matching between
`products and patients here.
`On the other hand, firms may reduce DTCA upon generic entry due to
`externality of advertising. In particular, many states in the U.S. have
`adopted mandatory substitution laws that require pharmacists to dispense
`generics if they are available unless doctors say otherwise.9 This means that
`even when patients request a branded drug that they saw on television and a
`physician actually prescribes it, the brand-name drug may not be dispensed
`at a pharmacy if equivalent generic drugs are available. Since the return from
`advertising may be lower under this situation, firms may reduce advertising
`expenditure upon generic entry.
`
`III(ii). Market-Expanding vs. Business-Stealing
`
`Next, I examine whether DTCA is used to expand the size of the market or
`shift market shares among existing brands. Advertising is viewed as market
`expanding when it purely increases the total size of the market. In contrast, it
`is referred to as business stealing when it solely shifts market share among
`existing brands. This distinction has long been recognized in advertising
`literature. Friedman [1983], for example, described the two types of advert-
`ising as ‘cooperative’ and ‘predatory,’ respectively. Roberts and Samuelson
`[1988] empirically examined the nature of cigarette advertising and found
`that it had primarily a market-expanding effect rather than a business-
`
`7 See Caves, Whinston, and Hurwitz [1991], Grabowski and Vernon [1992], Frank and
`Salkever [1997], and Wiggins and Maness [forthcoming] for the debate.
`8 For example, during the first week of patent expiration of Prozac, a blockbuster anti-
`depressant drug selling $2.6 billion in 2000, eighty percent of U.S. patients switched to a generic
`equivalent (The Financial Times, ‘Majority of Prozac-Users Switch to Generics,’ August 21,
`2001)
`9 See Hellerstein [1998] for more about the mandatory substitution law and its effects.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING OF PRESCRIPTION DRUGS
`
`357
`
`stealing effect. In contrast, Gasmi et al. [1992] found that advertising in the
`carbonated soft-drink industry is primarily characterized as business
`stealing. Typically, advertising has been viewed as welfare reducing unless
`it increases total market demand.10
`These distinctions are also important in the current context. As discussed
`in Section II, one of the main arguments in favor of DTCA was that DTCA
`encourages patients to visit physicians’ offices and seek medical help.
`Because many chronic diseases, such as high cholesterol and diabetes, are
`seriously under-diagnosed and under-treated, they argue, DTCA could
`potentially improve welfare by informing future patients of the existence of
`treatments. In other words, proponents argue that DTCA has primarily a
`market-expanding effect rather than a business-stealing effect. Empirically,
`I distinguish the two effects by examining whether DTCA is targeted to
`currently treated or untreated patients. Naturally, market-expanding
`advertising should be sensitive to the number of currently untreated
`patients, while business-stealing advertising should respond to the number
`of currently treated patients.
`
`III(iii). Drug Quality and DTCA
`
`Another issue of interest is the relationship between product quality and
`DTCA. Specifically, I ask whether firms spend more advertising dollars for
`high-quality drugs or low-quality ones. Traditionally, economists have
`discussed whether product quality and advertising are complements or
`substitutes. The most often cited theory that connects advertising to product
`quality is Nelson’s [1974] theory of advertising as a signal of quality. He
`argues that the mere fact that firms spend a lot of money in advertising
`reveals its high quality even when ads do not contain any explicit quality
`information. This is possible because, for experience goods whose quality
`can be judged only after consumption, high-quality products are more likely
`to attract repeat purchases. The return from advertising that induces initial
`purchases is higher for high-quality products, and thus high-quality firms
`will spend more on advertising. Milgrom and Roberts [1986] formalized
`Nelson’s idea by allowing both price and dissipative advertising to be used as
`signals of quality. They show that a separating equilibrium exists in which
`only high-quality firms advertise, as long as the marginal cost advantage of
`low-quality firms is not substantially large.11 If otherwise, price alone can
`signal quality and advertising will not be used as a signal.
`There are other situations, however, in which firms may advertise more
`when the drug is of high quality. In particular, if advertising can directly
`
`10 See Becker and Murphy [1993] (pp. 959–60) for more discussions.
`11 While I do not observe marginal costs in my data, marginal costs are typically small in the
`case of prescription drugs and may not increase substantially even for high-quality drugs.
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`358
`
`TOSHIAKI IIZUKA
`
`inform the quality of the product to consumers and marginal costs of high-
`quality products are not substantially higher than that of low-quality
`products, then marginal return from advertising may be higher for high-
`quality drugs. This would also encourage high quality producers to advertise
`more. In fact, this is also a plausible scenario in the current case since, as
`discussed before, prescription drug advertising is likely to convey credible
`quality information. Also, marginal costs are generally small and are not
`likely to increase substantially even for high quality drugs. Thus, at least two
`views predict that high-quality drugs will advertise more.
`
`III(iv). Order of Entry
`
`Next, I examine whether order of entry in each market affects the use of
`DTCA. This is certainly possible if the marginal return of DTCA is different
`depending on the timing of entry. One can easily think of a case in which
`early entrants advertise more than late entrants. For example, because the
`cost of learning is high for physicians, physicians might form a ‘habit’ and
`keep prescribing the same drugs, most likely pioneer drugs. In fact, such
`persistence of doctors’ prescription behavior has been shown in recent
`literature including Hellerstein [1998], Stern and Trajtenberg [1998], and
`Coscelli [2000]. Under this circumstance, return from advertising may be
`higher for early entrants, and this would make them advertise more than late
`entrants.
`Whether there is an asymmetry in the effectiveness of marketing
`instruments is a recent research agenda in marketing literature. Bowman
`and Gatignon [1996], for example, examined whether order of entry affects
`the effectiveness of advertising. Their results from two durables and three
`nondurables, however, did not support an asymmetric effect of advertising.
`In contrast, Shankar et al.
`[1998] showed that, using data from 13
`pharmaceutical brands in the 1970s and 1980s, noninnovative late entrants
`have less effective marketing spending compared to pioneers.
`It should be noted that if the order of entry does indeed affect the extent of
`advertising, then DTCA might affect market structure and firms’ R&D
`decisions in the long run. In particular, if early entrants enjoy the benefits of
`DTCA more than late entrants do, then returns for a pioneer would increase
`while the incentives to develop ‘me-too’ drugs would decrease. Thus, the race
`to become a pioneer may become intense and only a smaller number of firms
`may be able to exist in each market. DTCA may lead to a more concentrated
`market structure.
`
`III(v). Drug Age and DTCA
`
`Finally, I examine whether the use of DTCA varies depending on the age of
`the drug. Various advertising models predict different time paths. For
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`DIRECT-TO-CONSUMER ADVERTISING OF PRESCRIPTION DRUGS
`
`359
`
`example, if the role of advertising were to inform consumers about the
`existence of products as in Grossman and Shapiro [1984], then newer drugs
`would be advertised more often than older drugs. Over time, as people learn
`about the drug through advertisement, the informative role of advertising
`becomes less important, and thus DTCA may decrease as the drug gets
`older, other things equal. Nelson’s signaling model (e.g., Milgrom and
`Roberts [1986])Fwhich assumes perfect learningFpredicts the same time
`path because the return from signaling quality will diminish over time as the
`number of experienced consumers increases. Horstmann and MacDonald
`[1994], however, show that the conclusion may be reversed, i.e., advertising
`expenditures increase over time, if the learning of consumers is imperfect. In
`a separate paper (Horstmann and MacDonald [2003]), they examine
`compact disc players and show that advertising expenditures increase over
`time in this market.
`
`IV. DATA AND VARIABLES
`
`The data set was compiled from several sources, as described below. I have a
`total of 606 drug-year observations for 169 unique brand-name drugs over
`the period 1996–1999. These drugs belong to one of the following broad
`categories: central nervous system agents, respiratory agents, and renal and
`genitourinary agents.12 Drug Facts and Comparisons, a standard medical
`reference, was consulted to discover drugs that belong to each of these
`categories. Drugs approved before 1982 were excluded from the estimation
`because of the lack of comparable information. However, I included these
`drugs when counting the number of competing drugs in each therapeutic
`class (see below). Definitions of variables and data sources are summarized
`in Table I.
`DTCA Expenditure for each brand-name drug was obtained from TNS
`Media Intelligence/Competitive Media Reporting (CMR). CMR monitors
`advertising units and expenditures for several different media, including
`cable TV, network TV, newspapers and magazines. All ads for prescription
`drugs that appeared in these media are included in the CMR database. In the
`estimation, I use annual total DTCA expenditure as the dependent variable.
`Age of Drug was calculated as the year since FDA approval. The date of
`FDA approval was obtained from the FDA’s Orange Book. While the FDA
`approval date may not be exactly the same as the product launch date, the
`difference is usually not very large. Thus, I use the FDA approval date to
`calculate the age of the drug.
`
`12 I limit my samples to these categories largely due to the high cost of constructing the data
`set. While the estimation results may or may not extend to the remaining categories, these
`categories represent roughly 46% of industry sales in 1996 (S&P [1999]).
`
`r Blackwell Publishing Ltd. 2004.
`
`

`

`360
`
`TOSHIAKI IIZUKA
`
`Table I
`Definition of theVariables and Data Source
`
`Variable
`
`DTC ads
`HIGH_Q
`
`1st Move
`
`2nd Move
`
`AGE
`G_ENTRY
`
`THRP_COMP
`
`PTNT_SIZE
`
`CRNT_SIZE
`
`ACUTE
`
`IN