TCHICAGO
`
`
`Jr"
`
`'1' II Ii
`
`1.
`
`in. W 5 (1 H0012.
`
`Q THE UNIVERSITY HF EHIL‘AEI] PRESS lEIIJHNAlE
`
`The Booth School of Business. University of Chicago
`
`Does Intellectual Property Restrict Output? An Analysis of Pharmaceutical Markets
`
`Author(s): Darius Lakdawalla and Tomas Philipson
`
`Source: The Journal 01" Law & Economics, Vol. 55, N0.
`
`1 (February 2012), pp. 151-187
`
`Published by: The University of Chicago Press for The Booth School of Business,
`
`University of Chicago and The University of Chicago Law School
`Stable URL: http=//www.jstor.org/stable/10.1086/6633‘15
`Accessed: 14-12-2017 19:14 UTC
`
`LTSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
`
`range of content in a trusted digital archive. We use information technology and tools to increase productivity and
`
`facilitate new forms of scholarship. For more information about TSTOR, please contact support@jstor.org.
`
`Your use of the TSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
`
`http://about.jstor.org/terms
`
`The University of Chicago Law School, The Booth School of Business, University of
`Chicago, The University of’ Chicago Press are collaborating with ISTOR to digitize,
`preserve and extend access to The Journal of Law & Economics
`
`
`
`J STOR
`
`This content downloaded from 430.121.19 on Thu, 14 Dec 2017 19: 14:29 UTC
`All use subject to http://ab0ut.jstor.org/terms
`
`Argentum Pharm. LLC V. Alcon Research, Ltd.
`Case lPR2017-01053
`
`ALCON 2076
`
`

`

`Does Intellectual Property Restrict Output?
`An Analysis of Pharmaceutical Markets
`
`Darius Lakdawalla University of Southern California
`Tomas Philipson University of Chicago
`
`Abstract
`
`Standard analysis of intellectual property focuses on the balance between in-
`centives for research and the welfare costs of restraining output through mo-
`nopoly pricing. We present evidence from the pharmaceutical industry that
`output often fails to rise after patent expirations. Patents restrict output by
`allowing monopoly pricing but may also boost output and welfare by improving
`incentives for marketing, a form of nonprice competition. We analyze how
`nonprice factors such as marketing mitigate and even offset the costs of mo-
`nopoly associated with intellectual property. Empirical analysis of pharmaceu-
`tical patents suggests that, in the short run, patent expirations reduce output
`and consumer welfare by decreasing marketing. In the long run, patent expi-
`rations benefit consumers, but by 30 percent less than would be implied by the
`reduction in price alone. Focusing only on the pricing issues of intellectual
`property may lead to incomplete or even inaccurate conclusions for welfare.
`
`1. Introduction
`
`Intellectual property (IP) spurs innovation by increasing the rewards for dis-
`covery, but it does so by granting a monopoly in the event of discovery. According
`to standard analysis (see Nordhaus 1969), the research and development (R&D)
`benefits of a patent system must be weighed against the associated output lost
`to patent monopolies, which reduce price competition. This analysis implies that
`patent expirations always lead to increased competition, lower prices, and higher
`market output. From this point of view, Figure 1 is surprising. The figure depicts
`
`For helpful comments and suggestions, the authors would like to thank Jay Bhattacharya, Richard
`Manning, and Richard Wang along with seminar participants at the University of Chicago, University
`of California, Los Angeles, the National Bureau of Economic Research (NBER) Health Care Program
`meetings, the NBER Summer Institute, and the Southern Economic Association meetings. We are
`grateful to the National Institute on Aging (Lakdawalla) and the George J. Stigler Center at the
`University of Chicago (Philipson) for financial support.
`
`[Journal of Law and Economics, vol. 55 (February 2012)]
`䉷 2012 by The University of Chicago. All rights reserved. 0022-2186/2012/5501-0006$10.00
`
`151
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`152
`
`The Journal of LAW& ECONOMICS
`
`Figure 1. Effect of patent expirations on quantity sold, for a sample of prescription drugs
`
`the percentage change in quantity—comparing the month before patent expi-
`ration with the month after—for a sample of U.S. pharmaceutical products whose
`patents expired between 1992 and 2002.1 For about 40 percent of drugs, output
`falls after patent expiration and expands only modestly for many others.
`Figure 1 suggests that there may be more to a patent expiration than the end
`of monopoly pricing alone, and consequently more to the welfare effects of IP
`protection. We argue that the standard analysis of IP must incorporate various
`aspects of nonprice competition, which may reinforce or mitigate the effects of
`monopoly pricing. For example, while monopolists have incentives to restrict
`quantity through higher prices, they may also have different incentives to pro-
`mote their product through advertising, to provide durability of goods, and to
`vertically integrate with upstream or downstream firms. These forms of nonprice
`competition can change the efficiency impact of IP regulations by either miti-
`gating or reinforcing the conventional effects on price competition.
`Motivated by this idea, we examine the effect of marketing—a particularly
`important form of nonprice competition—on the static and dynamic efficiency
`
`1 To be specific, Figure 1 shows the percentage decline or growth in prescriptions filled (in grams)
`between the month before and the month after expiration. More detail on the data is given in Section
`3.2.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`153
`
`Figure 2. Mean trends in price and quantity for fully advertised drugs
`
`of patents.2 Patent expirations decrease the private returns to marketing, which
`may be limited when goods are sold at marginal cost. As a result, expirations
`may actually reduce output if they decrease marketing effort by enough to offset
`the impact of price reductions. From a normative point of view, advertising by
`a monopolist is valuable because it moves output toward its efficient level.
`Notably, this is true even if advertising is purely persuasive and provides no
`valuable information to consumers.
`To assess the importance of these arguments more fully, we estimate the impact
`of marketing on welfare using patent expirations in the U.S. pharmaceuticals
`market between 1990 and 2003. This industry is a natural choice for empirical
`analysis of R&D and marketing because it is among the highest spending in-
`dustries in both categories. The industry spends approximately 15 percent of
`sales on marketing and 16 percent of sales on R&D.3 By comparison, about 2
`and 3 percent of U.S. gross domestic product are allocated to advertising and
`R&D, respectively.
`Figures 2 and 3 provide some illustrative data from the pharmaceutical in-
`
`2 Different forms of nonprice competition merit separate analyses. For example, monopoly has a
`range of possible effects on quality provision. Mussa and Rosen (1978) show that monopolists will
`overdifferentiate their product and induce a lower quality choice by consumers. Subsequent authors
`have demonstrated how these results can be altered or even reversed under different specifications
`for demand (Gabszewicz and Wauthy 2002).
`3 Many drugs have seen dramatic increases in direct-to-consumer (DTC) advertising since the
`change in Food and Drug Administration guidelines on such advertising in 1997.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`154
`
`The Journal of LAW& ECONOMICS
`
`Figure 3. Mean trends in price and quantity for drugs not fully advertised
`
`dustry. The figures are based on data—described in Section 3.2—from 101 mol-
`ecules with expiring patents. Each figure depicts monthly time series, relative to
`the month of patent expiration, for branded quantity, branded price, total quan-
`tity, and total price, at the molecule level. Figure 2 depicts these trends for
`molecules that are advertised, while Figure 3 does so for molecules that are not.4
`For all drugs, price declines steadily after the month of patent expiration. For
`the nonadvertised drugs, quantity rises fairly steadily over this period as well.
`However, for the advertised drugs, quantity appears flat after patent expiration.
`This suggests that patent expirations have different effects for advertised drugs
`than for their nonadvertised peers.
`To estimate these effects more formally, we use the timing of patent expirations
`as instruments for the price and incentives of a molecule. Changes in supply
`induced by patent expiration allow us to identify the demand for drugs as a
`function of both price and advertising effort. The estimated demand function
`implies that in the short run (the first 5 months), output decreases after patent
`expiration because the reduction in advertising more than offsets the reduction
`in price. This output loss is estimated to cost consumers roughly $1 million per
`month for each drug whose patent expires. Not until several years have elapsed
`
`4 Figures 2 and 3 show the percentage change between the month of patent expiration and the month
`shown on the x-axis. In all cases, price is per gram. A fully advertised drug has at least 1 month of
`nonzero samples dispensed, at least 1 month of nonzero promotional visits to doctors, and at least 1
`month of nonzero medical journal advertisements. Drugs that do not meet these criteria are considered
`to be not fully advertised. The figures are discussed in more detail in Section 3.3.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`155
`
`does the price effect dominate the reduction in advertising. In the long run,
`patent expiration increases quantity and benefits consumers, but the reduction
`in advertising reduces the total gain to consumers from patent expiration by
`about 30 percent. However, even from a long-run perspective, monopoly mar-
`keting provides benefits to consumers. We estimate that, even if advertising is
`purely persuasive and provides no valuable information, the value to consumers
`is roughly 20–25 percent of total monopoly revenue, roughly on par with the
`costs of marketing to firms. Therefore, even if firms did not benefit from mar-
`keting at all, it would be no worse than welfare neutral.
`Our project integrates a great deal of work that has separately considered
`advertising and intellectual property.5 Several papers have studied the unique
`aspects of pharmaceutical advertising: Rosenthal et al. (2002) study direct-to-
`consumer advertising, while Bhattacharya and Vogt (2003) consider how brand
`loyalty affects pricing under IP. In the economic analysis of IP, an equally ex-
`tensive literature tackles the question of how to generate efficient R&D effort.
`There is a large literature analyzing the effects and desirability of public inter-
`ventions affecting the speed of technological change.6 Less effort has been devoted
`to studying the joint problem of advertising and IP, even though the interaction
`between these two factors has many important normative and positive impli-
`cations, particularly for the marketing of pharmaceuticals in the United States.
`The paper proceeds as follows. Section 2 considers the impact of nonprice
`competition on the welfare effects of patents and outlines the full impact of
`patents on static and dynamic welfare. Section 3 estimates demand as a function
`of price and advertising and infers changes in welfare from marketing and patent
`expiration. Section 4 concludes and discusses future research.
`
`2. Marketing and Intellectual Property
`
`From a positive point of view, IP protection has ambiguous effects on quantity
`provided, depending on the strength of incentives to market. Normatively, the
`static costs (or benefits) of patents depend on whether patent protection moves
`quantity toward or past the efficient level.
`
`2.1. The Welfare Effects of Patents
`
`W
`W
`as the annual level of aggregate welfare (social surplus)
`and
`Define
`C
`M
`under monopoly and competitive provision of an invention, respectively. The
`
`5 Kaldor (1949) provides a seminal analysis of advertising along both positive and normative
`dimensions. Dixit and Norman (1978) and Telser (1962) provide an initial discussion of the meta-
`preference approach to welfare analysis of advertising developed formally and systematically by Becker
`and Murphy (1993). There are also summary treatments of advertising in Tirole (1988), Shapiro
`(1982), Schmalensee (1987), and Bagwell (2005).
`6 Representative treatments include Nordhaus (1969), Loury (1979), Wright (1983), Judd (1985),
`Gilbert and Shapiro (1990), Klemperer (1990), Horstman, MacDonald, and Slivinski (1985), Gallini
`(1992), Green and Scotchmer (1995), and Scotchmer (2004).
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`156
`
`The Journal of LAW& ECONOMICS
`
`net present value of welfare associated with a patent of length t years is then
`given by
`
`W(t) p n(t)W ⫹ [n(⬁) ⫺ n(t)]W .
`
`M
`
`C
`
`n(t)
`is the date-zero present value of a claim that pays $1 for t years. The
`Here,
`net present value of profits associated with this patent is given by
`p(t) p n(t)p ,M
`represents monopoly profits. To represent technological investment
`where
`pM
`induced by IP protection, define the increasing, differentiable, and strictly con-
`cave function
`as the probability of discovering an invention, as a function
`m(r)
`of R&D investment r. The privately optimal R&D associated with a patent of
`length t maximizes expected profits:
`r(t) p arg max m(r)p(t) ⫺ r.
`
`(1)
`
`r
`
`This implies that increases in the profits expected from discovery also stimulate
`marketing activity, because innovators expect greater rewards (Nordhaus 1969).
`Therefore, innovation is complementary with all investments that stimulate prof-
`its, including marketing.
`This level of R&D induces the expected social surplus:
`ES(t) p m[r(t)]W(t) ⫺ r(t).
`
`(2)
`
`t
`
`r
`
`t
`
`The dynamically optimal patent length that maximizes expected welfare is there-
`fore given by the following first-order necessary condition:
`r [mW(t) ⫺ 1] ≥ m(⫺W ).
`The marginal gains from increasing R&D levels through IP (the left-hand side)
`are made up of the extra R&D induced by the patent extension,
`, multiplied
`rt
`mW(t) ⫺ 1
`by the net social value of that extra R&D,
`, which consists of the
`r
`marginal social gain from more invention net of research spending. The optimal
`patent life equates this marginal benefit of an extension with the marginal cost
`of the extension, which is the welfare cost of an additional year of monopoly
`(on the right-hand side). The marginal cost of patent expiration—the loss of
`welfare once the technology has been discovered—is given by the static welfare
`effect,
`
`(3)
`
`dW dnp (S ⫹ p ⫺ S ),
`dt
`dt
`
`M
`
`C
`
`M
`
`where SM and SC are consumer surplus under monopoly and competition, re-
`S ⫹ p ! S
`spectively. Patents are costly on the margin whenever
`; this is the
`M
`M
`C
`condition for static deadweight loss from monopoly. Below we demonstrate that
`marketing lowers the relative cost of patents and that it can sometimes lower
`cost to zero or below. In such cases, infinite patent length is desirable.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`157
`
`2.2. Positive Effects of Advertising under Patent Monopoly
`
`We first show that advertising limits and sometimes fully offsets the quantity-
`restricting effects of patent monopoly. Consider the standard monopoly profit-
`maximization model:7
`
`max Q(p, A)p ⫺ MC # Q(p, A) ⫺ A.
`
`p, A
`
`For simplicity, consider the constant elasticity demand function, Q(p, A) p
`␧ ⫺gA p
`. Monopoly equilibrium with advertising exists and is well defined when
`0 ! ␧ ! 1
`.8 The
`g 1 1
`demand is elastic to price and inelastic to advertising:
`and
`optimal price is given by the standard Lerner markup condition:
`
`[p]:p p
`
`cg
`g ⫺ 1
`
`.
`
`The first-order condition for optimal advertising equates the marginal value of
`marketing to its marginal cost, according to
`
`[A]:
`
`dQ
`dA
`
`(p ⫺ c) p 1.
`
`This expression demonstrates why perfectly competitive firms have no incentives
`to advertise. Without a markup, it is not valuable to stimulate more quantity.
`Using the constant elasticity form, the first-order condition for advertising
`can be rewritten as
`
`⫺g
`
`␧p (p ⫺ c) p A .
`
`1⫺␧
`
`Combining the two first-order conditions, the equilibrium level of advertising
`1⫺␧A p ␧c [1/ (g ⫺ 1)]
`is given by
`. The monopoly quantity can be written as
`
`Q p c
`
`M
`
`⫺g
`
`⫺g
`
`(
`
`g
`g ⫺ 1
`
`)
`
`␧
`
`A p c
`
`⫺g
`
`(
`
`g
`g ⫺ 1
`
`)
`
`⫺g
`
`[
`
`(
`
`␧c
`
`)(
`
`cg
`1
`g ⫺ 1 g ⫺ 1
`Q p c
`. And the
`By comparison, the competitive level of quantity is given by
`Q p c
`[g/ (g ⫺
`⫺g
`monopoly quantity in the absence of marketing would be
`⫺g
`M0
`M
`Q ! Q
`. It will always be true that
`and that marketing leads to higher
`1)]
`
`⫺g ␧/(1⫺␧)
`
`]
`
`)
`
`C
`
`M0
`
`.
`
`⫺g
`
`7 The static model is useful and appropriate for advertising. However, a more dynamic approach
`may be needed for the analysis of other types of nonprice competition, like quality, that influence
`research and development.
`8 The condition on the price elasticity is standard. The condition on the advertising elasticity
`follows as a corollary of the Dorfman-Steiner theorem. Observe that
`
`␧
`c
`A
`cq
`p p pq ⫺ cq ⫺ A p pq 1 ⫺ ⫺ p pq 1 ⫺ ⫺ p pq
`pq
`pq
`p
`g
`
`(
`
`)
`
`(
`
`)
`
`(1 ⫺ ␧)
`g
`
`.
`
`Nonnegative profits imply that
`
`␧ ! 1
`.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`158
`
`The Journal of LAW& ECONOMICS
`
`g
`
`MQ
`quantity provision in the marketplace.9 However,
`may be higher or lower
`CQ
`than
`, depending on the configuration of parameters.
`Q /Q p [g/ (g ⫺ 1)] A p (c/p) A
`⫺g
`␧
`␧
`M C
`Observe that
`. The higher the price
`markup, the greater the restriction in quantity; this is the standard incentive
`effect of monopoly, encapsulated by the first term. However, monopolists restrict
`(␧)
`quantity less when the responsiveness of demand to marketing
`is higher.
`To take a few concrete examples, monopoly quantity is nearly 50 percent
`,c p .1 g p 1.6
`
`higher than competitive quantity for the parameters
`, and
`␧ p .9
`; under this scenario, more than half of revenue (.9/1.6) is spent on
`
`,c p .1 g p 1.6
`␧ p .5
`marketing. In contrast, for
`, and
`, monopoly quantity is
`more than 80 percent below competitive quantity, and less than one-third of
`revenue (.5/1.6) is spent on marketing.
`This implies that marketing can partially or completely offset monopoly pric-
`ing and can even overcorrect for the quantity distortions of monopoly, depending
`on the strength of incentives to market. Note that the analysis so far is strictly
`positive in nature, as the competitive level of quantity may be equal to or below
`the efficient level.
`
`2.3. Normative Analysis of Patents with Advertising
`
`The cost of monopoly is the reduction in quantity suffered by consumers. In
`the absence of advertising, this is easy to calculate: the quantity provided after
`patent expiration is assumed to be the competitive and efficient level; the dif-
`ference between the monopoly quantity and the postexpiration quantity yields
`the social cost.
`The introduction of marketing makes the situation more complex for two
`reasons. First, competition may increase or decrease quantity. Second, since
`competition may not produce the efficient level of quantity, the competitive level
`of output cannot be used as a simple benchmark for efficiency. Nonetheless, the
`competitive level of quantity can usually help bound the welfare costs of patents,
`as we show.
`In general, the welfare effects of patents are a priori ambiguous because their
`quantity effects are ambiguous. Recall from Section 2.2 that patent expiration
`has offsetting effects on quantity. Patent expiration increases quantity by lowering
`the market price but decreases it by reducing market advertising. Ultimately, the
`welfare effects depend on the value of changes in quantity and (sometimes) on
`the direct consumption value of changes in advertising. The ambiguous effect
`on quantity thus creates ambiguity for the welfare effects of patents. Moreover,
`the same forces that move quantity in opposite directions—namely, price cuts
`and advertising reductions—also move consumer welfare in opposite directions.
`Therefore, patents may help or harm consumers, depending on the shape of the
`demand and cost curves of a particular industry.
`
`9 Suppose not. In this case,
`cannot be an equilibrium.
`
`␧A ! 1
`
`, which implies that
`
`A ! 1
`
`and that, on the margin,
`
`D ! 0
`A
`
`. This
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`159
`
`Figure 4. Welfare effects of patent expiration under informational advertising
`
`2.3.1. Advertising as Information
`
`We first consider the welfare effects of patents when advertising provides
`valuable information about a product but does not add value to the product
`itself or provide direct utility to consumers. In this case, advertising does not
`affect the true value of a good to consumers but does move perceived value
`toward the true value. Let
`represent the inverse demand as a function
`p(x, a)
`of quantity
`and advertising
`(x)
`. Price falls in quantity but rises in advertising.
`(a)
`We denote by
`the full-information demand curve defined by
`p(x)
`lim p(x, a) p p(x), Gx.
`ar⬁
`
`The change in welfare due to patent expiration is given by the change in true
`social surplus, which is evaluated at the true, fully informed demand curve. We
`define this as
`
`M 冕
`
`Info
`
`C
`
`D { S ⫺ S p p(q)dq ⫺ (p x ⫺ p x ).
`
`x C
`
`C C
`
`M M
`
`(4)
`
`xM
`
`Figure 4 illustrates this argument for the case in which patent expiration reduces
`quantity. The change in quantity is evaluated along the true demand curve,
`which differs from the observed demand curves both before and after patent
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`160
`
`The Journal of LAW& ECONOMICS
`
`expiration. In this case, the welfare cost of patent expiration is given by the area
`L, which measures the welfare loss from patent expiration, assuming zero mar-
`ginal cost of production, and yields the consumer surplus associated with the
`additional quantity consumed under monopoly.
`Since advertising moves observed demand toward the true demand curve, we
`can use observed consumer surplus as a lower bound on the true consumer
`surplus, according to
`
`冕
`
`冕
`
`x
`
`C
`
`p(q)dq ≥
`
`x
`
`C
`
`p(q, a )dq.
`M
`
`x
`
`M
`
`x
`
`M
`
`On the basis of this inequality, our empirical analysis uses the observed change
`in consumer welfare as a bound on the true change in welfare. In particular, we
`construct the estimator
`
`Info 冕
`
`˜D { p(q, a )dq ⫺ (p x ⫺ p x ),
`
`x C
`
`M
`
`C C
`
`M M
`
`(5)
`
`≤ D
`˜D
`if and only if patent expiration increases quantity. Therefore,
`where
`Info
`Info
`the estimator is a lower bound, in absolute value, on the true increase in welfare.
`
`xM
`
`2.3.2. Advertising as Persuasion
`
`Some forms of advertising seek to persuade rather than inform. An example
`is the provision of in-kind prescribing incentives to physicians. Informational
`advertising moves demand toward its fully informed level. Purely persuasive
`advertising moves demand above its true level and may create socially excessive
`consumption.10 To separate this case from that of advertising as consumption,
`suppose further that advertising confers no direct consumption benefits on phy-
`sicians or patients.
`The welfare effects of persuasive advertising depend on the strength of mar-
`keting incentives and whether these boost quantity past its efficient level. While
`persuasive advertising always pushes the demand curve above its efficient level,
`it does not always push equilibrium quantity past this point. Figure 5 illustrates
`this case, in which market demand, DM, exceeds the true demand curve, DT.
`Region L shows welfare loss due to patent monopoly with persuasive advertising,
`(cid:4)
`and L show this loss for patent monopoly without
`and the combined regions L
`advertising. Even though demand is pushed past its efficient level, the monopolist
`does not choose to boost demand by so much that the equilibrium quantity
`exceeds its efficient level, XC. As a result, the “inefficient” growth in demand
`actually reduces deadweight loss due to underutilization. Therefore, even per-
`suasive advertising partially offsets the monopoly restriction on quantity and
`thus improves social welfare.
`
`10 If advertising is only partially persuasive and fails to increase demand above its true level, the
`analysis is substantially similar to the case of advertising as information.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`161
`
`Figure 5. Welfare effects of patent expiration under persuasive advertising
`
`Implicit in this last point is the premise that advertising cost must be strictly
`less than the additional consumer surplus generated by the growth in utilization.
`This is a straightforward implication of profit-maximizing behavior by the mo-
`nopolist, who will never spend more than the incremental value created for
`consumers. An alternative case is one in which demand rises by so much as to
`push equilibrium quantity past XC. In this case, advertising may create static
`welfare loss because of overutilization.
`Empirically, the only difference with persuasive advertising is that the com-
`petitive demand curve coincides with the true demand and lies below the mo-
`nopoly demand curve. Therefore, the competitive demand curve serves as the
`best approximation of true demand.
`In this case, we use the estimator
`
`x C
`
`Pers 冕
`Info 冕
`
`˜D { p(q, a )dq ⫺ (p x ⫺ p x ).
`
`C
`
`C C
`
`M M
`
`(6)
`
`Observe that
`
`xM
`
`˜
`˜
`D p D ⫹
`
`Pers
`
`x C
`
`[p(q, a ) ⫺ p(q, a )]dq.
`
`M
`
`C
`
`xM
`˜
`˜
`! D
`. Intuitively, when we assume that
`The latter expression implies that
`D
`Pers
`Info
`advertising is persuasive, we are implicitly assuming that the true demand for
`the product is lower than in the informational advertising case. Under persuasive
`advertising, therefore, patent expiration benefits consumers less. As before, this
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`162
`
`The Journal of LAW& ECONOMICS
`
`estimator is a lower bound on the true absolute change in welfare due to patent
`expiration.
`
`2.3.3. Advertising as Consumption
`
`Suppose that advertising confers utility through two channels. The first is
`direct: exposure to advertising produces utility. For example, pharmaceutical
`companies may provide perquisites to consumers (in the form of samples) or
`to physicians (in the form of gifts or in-kind transfers). The second operates
`through complementarity with consumption. Advertising may increase the true
`value of and consumers’ willingness to pay for a product. That is, consumers
`may derive more utility from using a heavily advertised product. In this case,
`the consumer welfare effect of patent expiration satisfies
`
`D { S ⫺ S p V(a ) ⫹
`
`C
`
`C
`
`M
`
`冕
`
`x C
`
`p(q, a )dq ⫺ V(a )
`
`M
`
`C
`
`C
`
`冕
`
`0
`
`xM
`
`(7)
`
`p(q, a )dq ⫺ (p x ⫺ p x ).
`
`C C
`
`M M
`
`M
`
`⫺
`
`0
`
`V(a )
`V(a )
`represent the direct utility value of competitive and
`and
`The terms
`M
`C
`monopoly advertising levels, respectively. When advertising has consumption
`value, patent expiration can raise output while still lowering welfare: the decline
`in price raises output and welfare, but the reduction in advertising has a direct
`negative effect on welfare.
`These results are illustrated by Figure 6, which depicts the change in gross
`surplus that occurs at patent expiration when advertising provides utility. In that
`case, patent expiration lowers price and shifts demand inward. Regions G and
`L show the respective gain and loss in gross social surplus attributable to a
`simultaneous reduction in advertising and price. The welfare impact is ambig-
`uous and depends on the sizes of G and L. When advertising has value in itself,
`therefore, care must be taken when inferring changes in welfare from changes
`in output. For example, it is possible that the optimal patent life is infinite, even
`when patent expiration increases output.
`Another possibility is differential marketing to consumers with different will-
`ingness to pay. While price discrimination may be difficult, discrimination
`through marketing is much easier. This applies to the promotion of drugs to
`doctors, called “detailing” in pharmaceutical markets. Differential advertising
`across doctors and markets may act as a form of price discrimination. Since
`advertising cannot be resold, it is more easily implemented than traditional forms
`of price discrimination. Thus, advertising may shrink pricing inefficiencies and
`thereby lower the marginal cost of patent extension. Discriminatory advertising
`may lower or even remove the deadweight losses associated with patent mono-
`polies.
`When advertising has consumption value, it is necessary to estimate the direct
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`Intellectual Property and Pharmaceutical Markets
`
`163
`
`Figure 6. Welfare effects of patent expiration under advertising as consumption
`
`utility of advertising in order to capture the full value of patent expiration. In
`the absence of this estimate, we can say that consumers benefit less from patent
`expiration whenever they derive consumption value from advertising.
`
`3. Empirical Analysis
`
`This section investigates the empirical impact of pharmaceutical marketing
`on consumer welfare. Our approach is to use patent expirations as a means of
`identifying the demand curve for pharmaceuticals, where demand depends on
`both price and advertising effort. These estimates are then used to calculate how
`much patent expiration benefits (or costs) consumers.
`We focus on direct-to-physician marketing, which accounts for about 86 per-
`cent of all pharmaceutical marketing (Kaiser Family Foundation 2003). We es-
`timate the value of marketing under the alternative models of advertising as
`information (Section 2.3.1) and advertising as persuasion (Section 2.3.2). We do
`not have the data necessary to estimate the direct utility value of such marketing
`to consumers or physicians. If this exists, it will further increase the value of
`marketing and reduce the cost of patents.
`We begin by presenting our empirical model and approach to welfare esti-
`mation. We then describe our data and present descriptive analyses of the re-
`lationships between patent expiration, quantity changes, and marketing effort.
`Next, we discuss our approach to measuring advertising and lay out our iden-
`tification strategy. We finish with our estimated models and welfare effects.
`
`This content downloaded from 4.30.121.19 on Thu, 14 Dec 2017 19:14:29 UTC
`All use subject to http://about.jstor.org/terms
`
`

`

`164
`
`The Journal of LAW& ECONOMICS
`
`3.1. Model and Approach to Welfare Estimation
`
`The basic framework for this analysis is the following demand function:
`ln x p b ⫹ b ln p ⫹ b ln a ⫹ f ⫹ M(t) ⫹ ␧ .
`
`it
`
`0
`
`1
`
`it
`
`2
`
`it
`
`i
`
`it
`
`(8)
`
`x
`p
`is the corresponding
`is the price of molecule i in month t,
`In this equation,
`it
`it
`quantity of the molecule, and
`is a measure of advertising. There is also a
`a it
`molecule fixed effect,
`, and a polynomial time trend,
`. We are particularly
`M(t)
`f
`i
`interested in using the demand function to ascertain the effects of patent ex-
`piration on quantity and on welfare. It is straightforward to assess the quantity
`effects, but estimating the welfare changes (in terms of consumer surplus) re-
`quires more discussion.
`The demand function (and its associated inverse demand) implies forms for
`the changes in consumer surplus presented in Section 2.3. Consider first the
`cost of quantity restriction alone, which would be present without advertising.
`Suppose that
`represents demand at the consumer’s true valuation of the
`p(q)
`(cid:4)
`good. Monopoly quantity is given by
`. Finally, define by
`the counterfactual
`x
`x
`C
`M
`quantity that would obtain under competition if prices changed but advertising
`remained at its monopoly level. If
`dp/p
`is the percentage change in price due to
`(cid:4)
`x { x [1 ⫹ (dp/p) ␧]
`patent expiration, we can define
`.