The Unexpected Consequences of Asymmetric Competition.
`An Application to the Pharmaceutical Industry∗
`
`Carmine Ornaghi‡
`Micael Castanheira†
`Maria- ´Angeles de Frutos¶
`
`Georges Siotis§
`
`October 20, 2017
`
`Abstract
`
`This paper shows that an asymmetric competition shock that leads to a steep price
`drop in one market segment may benefit substitute products. Consumers move away
`from the cheaper product triggering a reverse competition effect. This result is driven by
`non-price competition: asymmetric shocks decrease some firms’ investment in promotion,
`which cripples their ability to lure consumers. We identify the conditions under which
`the lower priced product loses volume sales.
`
`To assess the empirical relevance of these findings, we study the effects of generic entry
`into the pharmaceutical industry. We exploit a large product-level dataset for the US
`covering the period 1994Q1 to 2003Q4 and find strong empirical support for the model’s
`theoretical predictions. Our estimates rationalize the observation that a molecule that
`loses patent protection (the originator drug plus its generic competitors) often experiences
`a drop in the quantity market share –despite being sold at a fraction of the original price.
`
`JEL Classification: D22, I11, L13
`Keywords: Asymmetric competition, Pharmaceutical industry, Generic entry
`
`∗We would like to thank Laurent Bouton, Guilhem Cassan, Christopher Cotton, Raffaele Fiocco, Margaret
`Kyle, Patrick Legros, Alessandro Lizzeri, Laurent Mathevet, Jacopo Perego, R´egis Renault, Patrick Rey,
`Pablo Querubin, Tobias Salz, Fiona Scott Morton, Denni Tommasi, and Philippe Weil, as well as seminar
`participants at Oxford University, ECARES, the Paris School of Economics, Queen’s University, Universidad
`Carlos III, Universit´e de Cergy-Pontoise, Universit´e de Lausanne, EARIE2016, and CRETE2016.
`†ECARES (Universit´e Libre de Bruxelles - SBS-EM) and CEPR. Micael Castanheira is “Directeur de
`recherche” FRS-FNRS and gratefully acknowledges their financial support.
`‡University of Southampton
`§Universidad Carlos III de Madrid and CEPR. Georges Siotis gratefully acknowledges the financial support
`from the Ministerio Econom´ıa y Competitividad (Spain) grants Beca I3 2006/04050/011, ECO2015-65204-P,
`MDM 2014-0431, and Comunidad de Madrid grant MadEco-CM (S2015/HUM-3444).
`¶Maria- ´Angeles passed away before being able to complete this project. We owe a lot to her inspiration
`and enthusiasm, and dedicate this paper to her memory.
`
`Exhibit 1087
`ARGENTUM
`IPR2017-01053
`
`000001
`
`

`

`1 Introduction
`
`To attract customers, firms selling imperfect substitutes do not only cut prices. They also
`
`invest in non-price instruments, such as advertising and brand management. In the presence
`
`of product differentiation and asymmetric competition shocks, the entry of a firm or the
`
`launch of a new product will not squeeze the profit margins of all incumbents in the same
`
`manner. Some experience intense margin compression, whereas others remain comparatively
`
`shielded.
`
`Our model shows that, if we overlook non-price competition, asymmetric and symmetric
`
`shocks always have similar effects: demand shifts towards the cheaper market segment (call
`
`it A). Instead, when we take account of the non-price dimension, asymmetric competition
`
`shocks may give rise to what we call a reverse competition effect.
`
`The reason is that intense competition also cripples firm A’s capacity to invest in non-price
`
`instruments. This produces an opposite shift in demand, beneficial to the more expensive
`
`product segment (call it B). Hence, when non-price instruments are an important determi-
`
`nant of market outcomes, and when one segment (but not the other) moves from imperfect to
`
`perfect competition, allocative efficiency can be reduced. This holds whether demand even-
`
`tually moves towards A or towards B. Further, even when non-price instruments do not
`
`directly enter consumers’ utility, their surplus may be reduced by competition. This only
`
`happens when the demand shifts strongly towards B.
`
`This is more than a theoretical construct: using a data trove tracking prices, promotion,
`
`and quantities sold, we show that competition by generics in the trillion-dollar pharmaceutical
`
`market often fails to put effective pressure on the drugs that remain protected by a patent.
`
`Despite price drops as high as 45% for the drug experiencing generic entry, it is often the
`
`market share of competing molecules that increases. The volume market share of the molecule
`
`that is now cheaper —the originator drug plus its chemically equivalent generic version—
`
`drops, on average, by 31% in the pharmacy channel and by 26% for drugs sold in hospitals.
`
`As we detail below, both our theoretical and empirical findings show that, quite counter-
`
`intuitively, this phenomenon is more pronounced when molecules are close substitutes and
`
`when market size is large.
`
`1
`
`Ex
`hi
`
`000002
`
`

`

`To analyze these effects formally, we propose a stylized model in which two firms, A
`
`and B, each produce a differentiated product. Consider, first, the situation in which they
`
`compete only in price. Following standard intuition, the more substitutable the goods, the
`
`lower initial prices will be. Then, firm A is confronted with the entry of a new competitor
`
`that sells a direct substitute for its product. Absent capacity constraints, this competition
`
`shock drives the price of A down to marginal cost and forces B to also react with a lower
`
`price. In this situation, “competition works as expected”.
`
`What happens when the two firms also rely on non-price instruments such as advertising
`
`to attract consumers? In this setting, the asymmetric competition shock experienced by A
`
`also induces it to cut down investment in the non-price instrument. Whenever the non-price
`
`shock dominates the price shock, B sees its residual demand expand.
`
`Under which conditions does this reverse competition effect materialize? Quite surpris-
`
`ingly, the problem is more acute when A’s and B’s products are closer substitutes. The
`
`reason is that the more substitutable the two goods are, the more aggressively A and B
`
`compete prior to the entry of the third firm (before generic entry in the case of the phar-
`
`maceutical industry). This translates into initially lower prices and higher “promotion” (the
`
`non-price instrument that we can measure in our data). In that situation, generic entry has
`
`a comparatively small impact on prices; the reduction in promotion dominates. High levels
`
`of differentiation have the opposite effect: prices are initially high and promotion low. Then,
`
`generic entry affects primarily prices: both A’s and B’s prices drop.
`
`The model also informs us on the expected effects of demand elasticity: we find that
`
`a lower price elasticity of demand increases the likelihood that B benefits from the stiffer
`
`competition faced by A. The same goes for market size: the reverse competition effect is
`
`more likely to hold in large markets because B maintains a high level of promotion.
`
`Clearly, the pharmaceutical industry is a particular one. Yet, it is also an ideal testing
`
`ground to assess these predictions. First, we can precisely disentangle asymmetric shocks,
`
`caused by the loss of exclusivity (LoE), from the entry of new competing products. Such a
`
`clear dichotomy between the launch of new products and the loss of market power for a single
`
`product is difficult to observe in other markets. Second, non-price competition is particularly
`
`important: for large players, promotion represents 15% to 20% of total sales, about the same
`
`2
`
`000003
`
`

`

`as R&D.1 Third, agency issues between patients, physicians, and insurances likely increase
`
`the sensitivity of demand to non-price instrument relative to prices, which magnifies the
`
`effects we are after. Fourth, we find that elasticities differ between hospitals and pharmacies.
`
`This allows us to test whether a lower price elasticity of demand indeed benefits B. Fifth,
`
`the market is economically relevant: worldwide sales totaled nearly a trillion US dollars in
`
`2013, while the US market stood at 374 billion dollars in 2014.2 Last, but not least, given
`
`the long time span between patent filling and Loss of Exclusivity (LoE), actual market size
`
`and the degree of substitutability of competing products cannot be predicted ahead of actual
`
`launch. This produces substantial exogenous variation across episodes of generic entry that
`
`we exploit in our regressions.
`
`We start with a sample that covers all prescription sales in the U.S. between 1994Q1
`
`and 2003Q4 (40 quarters). From that dataset, we extract all the therapeutic classes (“ATC3
`
`markets”) for which data on prices, quantities, and promotional efforts are available. We then
`
`crossed these data with that of the FDA to identify episodes of generic entry (see Section 4).
`
`This leaves us with 95 episodes of generic entry scattered over 53 different ATC3 markets.
`
`The size of this sample allows us to exploit the (heterogeneous but always large and
`
`asymmetric) shocks to competition associated with LoE to identify the coefficients of the
`
`demand function. As shown in Section 5.1, the price-to-promotion elasticity ratio is lower
`
`in the pharmacy channel than in the hospital channel. We use this difference to test–and
`
`confirm–our theoretical predictions relating these elasticities to the evolution of market shares.
`
`After controlling for other possible sources of heterogeneity, we find that, on average,
`
`generic entry alone causes a 12% increase in market share for molecules that remain on
`
`patent. The effect is smaller in the hospital channel: the higher price elasticity reduces the
`
`magnitude of this effect by about 3 percentage points. We also propose a novel measure
`
`1In the Oxford Handbook of the Biopharmaceutical Industry, Harrington (2012) estimates the R&D to
`be at 17.9% of total net sales for the period 2001-2005, and Kenkel and Mathios (2012) report that the
`advertising-to-sales ratio was 18% in 2005 in the U.S. As points of comparison, they highlight that, in 2010,
`advertising stood at 4.5% of total net sales for General Motors (a car producer), 9.5% for Anheuser Busch
`(a beer producer) and 10.8% for Kellogg (breakfast cereals). The figures are typically smaller for most other
`R&D-intensive industries. For instance, in 2013, Apple spent 3% of its total net sales on R&D and 0.4% on
`advertising (Apple 2013: 10-K SEC submission). See also Manchanda et al. (2005)
`2Source:
`http://www.statista.com/statistics/263102/pharmaceutical-market-worldwide-revenue-since-
`2001/ and http://www.statista.com/statistics/238689/us-total-expenditure-on-medicine/
`
`3
`
`000004
`
`

`

`of product differentiation for the pharmaceutical industry based on the number of modes of
`
`action within a therapeutic class. We argue that the existence of different modes of action to
`
`treat a particular condition is indicative of more differentiation. We find that differentiation
`
`knocks another 4 percentage points off the market share gain of the competitors. Finally,
`
`the market share gain is further reduced by 7 percentage points in “small” markets. Each of
`
`these observations is in line with the theoretical predictions sketched out above.
`
`Related literature.
`
`Our paper is at the intersection of several literature strands, includ-
`
`ing industrial organization, advertising, and health economics. With regard to our empirical
`
`application, the existing literature on competitive interactions in the pharmaceutical industry
`
`has produced a complex, and sometimes contradictory, picture. One group of papers analyzes
`
`inter-brand competition when drugs are still patent-protected (see, for instance, Brekke and
`
`Kuhn (2006) for a theoretical model and Dave and Safer (2012) for empirics). de Frutos, Or-
`
`naghi and Siotis (2013) analyze inter-brand competition when the proportion of brand-loyal
`
`consumers is endogenously determined by promotional effort.
`
`Another strand focuses on intra-molecular competition following loss of exclusivity — i.e.,
`
`when a generic bio-equivalent drug can legally come to market (e.g. Scott Morton (2000)).3
`
`It was in that context that the “generic entry paradox” has been unearthed (the paradox
`
`being that the price of the originator drug often goes up following the launch of a competing
`
`chemically equivalent generic). This empirical finding has been thoroughly documented (see
`
`a.o. Caves et al. (1991); Regan (2008); Vandoros and Kanavos (2013)).
`
`The few papers that attempted to simultaneously analyze pre- and post-LoE competition
`
`have produced a mixed picture. For instance, Stern (1996) provides evidence of intense inter-
`
`molecular competition, whereas Ellison et al. (1997) reports strong intra-molecular rivalry
`
`between the originator and the generic version of the drug, as well as weak (or insignificant)
`
`inter-molecular competition.
`
`A related literature focuses on the relative importance of the persuasive and informative
`
`roles of promotional effort (Ching and Ishihara (2012)) and on whether detailing and direct-
`
`3See Grabowski and Kyle (2007) for a description of generic entry in the US in the period 1995-2005
`and Berndt and Dubois (2016) for a comparison of generic penetration across OECD countries for the period
`2004-2010.
`
`4
`
`000005
`
`

`

`to-consumer advertising have a market expansion effect (Ching et al. (2016); Iizuka (2004,
`
`2005); Fischer and Albers (2010)). Another strand analyzes the effectiveness of promotional
`
`effort: Mizik and Jacobson (2004) estimate the long run effect of detailing and sampling
`
`on prescriptions for three drugs. Manchanda, Rossi and Chintagunta (2004) assess whether,
`
`from a business perspective, detailing is misallocated across individual physicians. Narayanan
`
`and Manchada (2009) show that the persuasive effect dominates at the end of a molecule’s
`
`exclusivity period.
`
`Huckfeldt and Knittel (2011) show that evergreening strategies (the launch of a second-
`
`generation product by the same originator) helps explain instances of volume market share
`
`drop of the previous generation molecule, despite being sold at a fraction of the original
`
`price. Lakdawalla and Philipson (2012) share our motivation (volume sales drop following
`
`LoE) and exploit a similar sample. The main difference lies in the fact that we explicitly
`
`model competition. This allows us to derive testable hypotheses regarding driving forces
`
`underpinning the evolution of post LoE volume market shares.
`
`The remainder of the paper is organized as follows. Section 2 presents some unexpected
`
`facts that spur the paper’s central research question. Section 3 presents the model and derives
`
`testable implications. Section 4 describes the data, while Section 5 presents the empirical
`
`results. Section 6 reports rubustness and sensitivity checks. Section 7 concludes by discussing
`
`how the presence of non-price instruments can lead to mismatch between consumers and goods
`
`when competition is asymmetric.
`
`2 Generic competition: some empirical regularities
`
`2.1 Price, patents, and quantities
`
`Once on the market, the life cycle of a patent-protected pharmaceutical drug can be broken
`
`in two distinct stages. The first covers the period spanning market launch until the firm loses
`
`exclusivity, which usually stems from patent expiry. During that phase, the producing firm
`
`has exclusive rights over the production and distribution of the drug and can exercise market
`
`power. The second phase begins after loss of exclusivity (LoE), when generic equivalents can
`
`enter the market to compete with the originator firm.
`
`5
`
`000006
`
`

`

`The introduction of a chemically equivalent competitor produces a dramatic change in
`
`competitive conditions. Generics are typically sold at a fraction of the price of the original
`
`brand and exert strong competitive pressure on the original branded product: Grabowski
`
`et al. (2014) show that, for branded drugs facing first generic entry in 2011-2012, brands
`
`retained, on average, only 16% of the molecule market after one year.
`
`Figure 1 provides another perspective on these evolutions. Lumping together the original
`
`product and its generic equivalents, it depicts the evolution of the mean and median price
`
`and quantity for the 95 molecules that experienced generic entry in our dataset (U.S. data
`
`for the period 1994-2003). Time is expressed in quarters, and we denote as “date 0” the
`quarter in which firms lose exclusivity. We normalize to 1 values at quarter −12.4 Before
`patent expiration (quarters −12 to 0), the price of the original molecule is slightly increasing.
`
`Figure 1: Price and Quantity around generic entry
`
`4We control for the fact that not all molecules are observed in all quarters (for instance, if a patent
`expires four quarters before the end of the data, we have only four data points after patent expiration) by: i)
`computing the price change over two consecutive periods for all available molecules; ii) computing the average
`of these variations for each quarter before and after patent expiration; and iii) constructing an index that
`starts at 1 and that varies over time following the average variations computed at stage (ii). We follow the
`same approach to compute the median price and all the other statistics in this graph.
`
`6
`
`000007
`
`

`

`Then, within a year of the loss of exclusivity (LoE), mean (respectively, median) prices drop
`
`by about 30% (20%). Within three years, the drop reaches about 50% (40%).5
`
`As Figure 1 makes clear, despite being sold at a fraction of the original price, the gener-
`
`icized molecule actually experiences a drop in volume after LoE (black curves): the average
`
`and median quantities drop by more than 25% three years after patent expiry. In other words,
`
`after LoE, the combined volumes of the branded and generic producers are substantially be-
`
`low the volume of the single branded drug when it is sold at a price embodying monopoly
`
`power.
`
`As our econometric analysis below shows, this means that generic entry mainly benefits
`
`competing molecules. This despite the fact that the latter barely adjust their prices. Put
`
`differently, few new patients are directed to the cheap genericized molecule, and a number of
`
`existing patients switch to competing molecules at the time that their initial treatment be-
`
`comes cheaper. Neither the rationales for the so-called generic entry paradox (cf. footnote 5)
`
`nor Third Party Payer reimbursement rules can explain why cross-price elasticities suddenly
`
`seem to take the “wrong sign”.
`
`2.2 Loss of exclusivity and promotion intensity
`
`Turning to non-price competition, Figure 2 shows that LoE also triggers a major drop in
`
`the firms’ promotional effort (we will use the terms detailing, promotion and advertising as
`
`synonymous). Using data from IMS-Health, we measure the firms’ drug-specific spending on
`
`personalized visits to general practitioners and hospital specialists, free samples dispensed to
`
`physicians, and advertising in professional journals. All these instruments affect the physi-
`
`cians’ incentives to prescribe one drug rather than another. The data reveal that promotion
`
`falls continuously over the 12-quarter period before patent expiration, with a sharp accelera-
`
`tion around the time of LoE. At time 0, promotion effort has dropped by 50%. Four quarters
`
`5Although the average price of the molecule (displayed) falls following generic entry, this is not always the
`case for the price of the branded drug (not separately depicted in Figure 1). Sometimes, the latter remains
`constant or even increases; this is the so-called generic entry paradox (for empirical evidence, see Regan (2008)
`for the U.S. and Vandoros and Kanavos (2013) for the EU). This behavior is usually attributed to the fact
`that a subset of patients insist on purchasing the brand, even at a higher price. This allows branded producers
`to keep extracting rents on a (shrinking) subset of patients.
`
`7
`
`000008
`
`

`

`Figure 2: Price and Promotion around generic entry
`
`later, the median drop is close to 95%. At 12 quarters after LoE, median spending is zero.6
`
`The fact that price and promotion pull demand in opposite directions was already empha-
`
`sized by Caves, Whinston and Hurwitz (1991), who observed that “generic entry brings with
`
`it two offsetting effects: first, generic entrants offer significantly lower prices, which tend to
`
`expand overall sales of the drug, but their arrival also leads to a significant reduction in the
`
`level of advertising for the drug, which acts to counterbalance this price effect”. However,
`
`they did not explore the matter further, either theoretically or empirically.
`
`Posterior empirical results have confirmed that total molecule sales can increase or de-
`
`crease after LoE. Berndt et al. (2003) and Lakdawalla et al. (2007) find that the market
`
`share of the molecule losing exclusivity experiences a fall. Aitken et al. (2009, 2013) find
`
`the opposite. Duflos and Lichtenberg (2012) find that “the net effect of patent expiration on
`
`drug utilization is zero”.
`
`6The average lies above the median because some molecules continue to be promoted. For instance,
`high levels of promotions are observed for Prozac (fluoxetine) because Eli Lilly & Co.
`introduced weekly
`delayed release capsules of the drug just before LoE in an attempt to stem the post-patent decrease in sales of
`their daily dosage. Similarly, we observe positive spending for Zantac (ranitidine) and Tagamet (cimetidine),
`probably because some of their lower-dosage tablets are available over-the-counter (no prescription required).
`
`8
`
`000009
`
`

`

`While it is intuitive that either of two opposite forces may dominate, the literature falls
`
`short of explaining when and why one or the other outcome should materialize. Our analysis
`
`of asymmetric competition shocks fills this gap: we can identify the precise market structures
`
`for which companies’ advertising strategy will produce a drop in total volume sales of the
`
`molecule losing exclusivity, and the complementary set of parameters that will lead to an
`
`increase. We can then test these theoretical implications on pharmaceutical data. The results
`
`allow us to attribute these differing evolutions to firms’ pricing and promotion strategies.
`
`3 The model
`
`To analyze the effects of asymmetric competition shocks, we propose to build on a textbook
`
`model of differentiated Bertrand competition with advertising. That is, (1) initially, two
`
`firms sell differentiated products and compete through price and non-price instruments. In
`
`contrast with the standard textbook approach, we consider (2) competition shocks that are
`
`highly asymmetric: the new entrant is a perfect substitute for one good in an otherwise
`
`horizontally differentiated industry. (3) In the spirit of Inderst and Ottaviani (2012), con-
`
`sumer decisions are mediated through intermediaries who can be persuaded to modify their
`
`purchasing recommendations. In our framework, however, this is only relevant for the nor-
`
`mative analysis: realized consumption decisions may differ from those that would maximize
`
`the utility of the final consumer.
`
`Formally, two firms, A and B, compete both in prices and in “advertising,” or some other
`
`type of fixed cost investment that stimulates demand. Starting from that initial situation, we
`
`study the effects of the entry of a third firm, G, that produces a perfect substitute for firm A’s
`
`product? Intuitively, substitutability implies that A’s advertising effort produces a positive
`
`spillover on G’s demand (the way that a brick and mortar store’s advertising for a given
`
`product also stimulates online demand for the same product). Under these circumstances,
`
`competition between A and G causes both the price and advertising effort of A to drop
`
`substantially. Our central question is: what happens to the demand for B?
`
`Shifting to a terminology adapted to the pharmaceutical industry, consider a market
`
`in which two firms’ molecules compete for physicians’ prescriptions. We derive the firms’
`
`9
`
`000010
`
`

`

`promotion and pricing strategy given the quality θJ of their molecule, the heterogeneity of
`
`treatment responses across patients, e,7 and the agents’ sensitivity to price, δ.8
`
`Consider the case of a patient who goes to her physician with a condition that requires
`
`treatment. The intrinsic utility of the Patient-Physician-Payer triplet (“P3” henceforth)9 i’s
`
`from using treatment A or B is, respectively (we introduce detailing below):
`
`A = θA − δpA + εi,
`U i
`B = θB − δpB − εi,
`U i
`
`where pJ is the price of molecule J ∈ {A, B}; εi ∼ U(cid:2)− e
`
`(1)
`
`(cid:3) is the relative efficiency of drug
`
`(2)
`
`2 , e
`2
`A as opposed to B to treat patient i’s condition; and U denotes the uniform distribution.10 A
`
`larger value of e implies that patients are more heterogeneous in their response to treatments
`
`A and B, and hence that the two molecules are more distant substitutes. δ accounts for the
`
`triplet’s (P3) sensitivity to price.
`
`Note that assumptions such as the linearity of the demand schedule and the uniform
`
`distribution of patients would be inappropriate if we were to estimate the model structurally.
`
`However, we do not have individual consumption or prescription data (see Section 4), and our
`
`questions are rather orthogonal to the ones that structural estimations are meant to address.
`
`We thus choose to simplify the model to focus on the market shares of each molecule and, in
`
`Section 5, carry out the empirical analysis on the same market variables to test the model’s
`
`main predictions.
`
`Focusing on market shares, we let patients with a sufficiently good fit with drug A (i.e.,
`
`7 These parameters directly relate to vertical and horizontal product differentiation in classical Industrial
`Organization analyses. However, product differentiation is not a choice variable in our model.
`8This model specification fits the industry’s description provided by Berndt (2002): “Within many ther-
`apeutic classes of drugs, a number of possible substitute medications exist, and in such cases, the market
`structure is more appropriately depicted by the differentiated product oligopoly framework. In such a set-
`ting, it is useful to envisage the optimal profit maximizing price as equaling marginal cost plus a positive
`margin, where the margin depends on benefits and attributes (including prices) the firm’s own drug relative
`to other drugs in the therapeutic class, on attributes of non-drug therapies, patient heterogeneity and other
`demand-side considerations.”
`9The “payer” can either be a Third Party Payer (TPP), patient out of pocket expenditure, or a combination
`of both.
`10Focusing on a single random variable ε that can be either positive or negative implicitly eliminates patients
`with negative valuations of the two molecules, who have no reason to consume either of the two drugs. Thus,
`issues of aggregate under- or over-prescription are beyond what this model can capture.
`
`10
`
`000011
`
`

`

`with εi sufficiently positive) buy treatment A, and all the others buy drug B.11 Letting µ
`
`denote market size (or the number of afflicted patients), we identify the patient i who is
`
`indifferent between A and B to determine quantities in the absence of detailing:
`
`(cid:18)
`
`1 − F
`
`(cid:20) ∆θB − δ∆pB
`(cid:21)
`(cid:20) ∆θB − δ∆pB
`
`(cid:21)(cid:19)
`
`× µ
`
`QA =
`
`QB = F
`
`2
`
`× µ,
`
`2
`where F represents the CDF of εi, ∆θB ≡ θB − θA, and ∆pB ≡ pB − pA. We associate
`drug A with the oldest molecule, while B is more recent: firm A loses exclusivity before B.
`For the sake of the argument, we focus on the case ∆θB ≥ 0, since more recent drugs (here:
`B) are expected to be more effective than older drugs (here: A). However, all the results
`
`extend to the complementary case of ∆θB < 0. Thus, when they cannot promote their drugs
`
`(superscript N D for the “No Detailing” case), the two firms’ respective profits are:
`
`(cid:20) 1
`(cid:20) 1
`
`2
`
`2
`
`πN D
`A
`
`πN D
`B
`
`= pA ×
`
`= pB ×
`
`(cid:21)
`(cid:21)
`
`− ∆θB − δ∆pB
`2e
`∆θB − δ∆pB
`2e
`
`+
`
`× µ,
`
`× µ,
`
`(3)
`
`(4)
`
`where the terms between brackets are, respectively, the market shares of A and B when we
`
`substitute for the value of F under the uniform distribution.12
`
`Generic Entry.
`
`This setup describes a duopoly market: each firm’s patent gives it
`
`11This model specification assumes that all the patients who require a treatment receive one (full market
`coverage). This requires that the quality θJ of both molecules is sufficiently high relative to the equilibrium
`prices–formally, θA + θB > 2e (see Appendix 1). To capture incomplete coverage, we used another model
`specification that adds the distribution of willingness to pay ωi into the utility function:
`
`A = θA − δpA + εi + ωi, and U iB = θB − δpB − εi + ωi.
`U i
`
`(cid:16)
`
`(cid:16)
`
`(cid:17)(cid:17)
`
`Solving for the demand for, say, A, then shows that QA in equation (3) must be multiplied by:
`1 − Fω
`δpA − θA − ∆θB−δ∆pB +e
`in the simple case where the distribution of ω is also uniform. This
`4
`term captures the market contraction/expansion effects of higher/lower prices. This does not affect the sub-
`stance of our analysis of market shares. Since we do not have a strong case as to whether there is initially
`under- or over-provision of drugs, we decided not to incorporate these effects in the model: they make the
`analysis a lot less tractable, for little benefit.
`12Formally:
`
`
`e = 12 + εe ,∀ε ∈ [−e/2; e/2]
`
`1,∀ε > e/2.
`and the PDF is f (ε) = 1/e, ∀ε ∈ [−e/2; e/2] .
`
`F (ε) =
`
`ε+e/2
`
` 0,∀ε < −e/2
`
`11
`
`000012
`
`

`

`exclusivity for the sale of its molecule. Here, we turn to the effects of A losing that exclusivity
`
`(LoE), while firm B retains its patent protection and monopoly power
`
`The first case we study is the one in which there is no detailing.
`
`In the absence of
`
`detailing, an equilibrium is characterized by a pair of prices in which firms maximize profits
`in (3) − (4). Loss of exclusivity implies that chemically equivalent generics can compete
`directly for A consumers. Based on the evidence (see a.o. Grabowski et al., 2014), we let
`
`competition among generic producers reduce the price of A down to marginal costs, which
`
`we normalize to zero without loss of generality.
`
`The post-generic-entry equilibrium is then characterized by the price that maximizes
`
`B’s profits when pA = 0. Unsurprisingly, the results in Appendix 1 show that generics
`
`competition for the A market can only drive down the price and market share of drug B. We
`
`also find that, in an interior equilibrium, price sensitivity determines the magnitude of the
`
`price reduction, but does not influence market shares.
`
`Promotion.
`
`As discussed in the introduction and in Section 2.2, the pharmaceutical in-
`
`dustry stands out for its high promotional intensity. Through their detailing and sampling
`
`activities, pharmaceutical companies devote substantial resources to inform physicians and
`
`provide them with a number of perquisites, sometimes contingent on their prescription be-
`
`havior. This non-price competition component is also dramatically modified upon generic
`
`entry: price competition amongst producers brings detailing down to 0.
`
`We assume that promotion is persuasive: it stimulates prescriptions without affecting the
`patient’s intrinsic utility (1) − (2) nor bringing fresh information to doctors. As we discuss
`below, this reflects the situation at the end of a molecule’s life cycle.
`Formally, when firm J spends C (aJ ) ≡ a2
`J /2 on detailing, it produces an autonomous
`increase in the demand for drug J from θJ to θ(cid:48)
`J = θJ + aJ . Given an action profile
`{aA, aB, pA, pB} , the resulting demands are then:
`
`(cid:21)(cid:19)
`
`× µ,
`
`2
`
`× µ,
`
`2
`
`12
`
`(cid:18)
`
`1 − F
`
`QD
`A =
`
`QD
`B = F
`
`(cid:20) ∆θB + ∆aB − δ∆pB
`(cid:21)
`(cid:20) ∆θB + ∆aB − δ∆pB
`
`000013
`
`

`

`where superscript D denotes Detailing and ∆aB ≡ aB − aA. The firms’ profits become:
`− ∆θB + ∆aB − δ (pB − pA)
`2e
`∆θB + ∆aB − δ (pB − pA)
`2e
`
`(cid:20) 1
`(cid:20) 1
`
`2
`
`2
`
`+
`
`A = pA ×
`πD
`B = pB ×
`πD
`
`,
`
`.
`
`× µ − a2
`
`A 2
`
`× µ − a2
`
`B 2
`
`(cid:21)
`(cid:21)
`
`Discussion of the main assumptions. We made two important assumptions. First, pa-
`
`tients do not observe the relative importance of price and promotion in determining their
`
`physician’s prescription decision. The fact that both prices and promotion may influence
`
`prescription behavior is well grounded in facts, and the purpose of the theoretical model
`
`by Inderst and Ottaviani (2012), who study how commissions and kickbacks –and the con-
`
`sumers’ information about them– influences eventual market outcomes and welfare. Iizuaka
`
`(2012) provides evidence that physicians respond to economic incentives in their prescription
`
`decisions. The association between payments to physicians and their prescription behavior
`
`can also be assessed on the basis of publicly available data (Grochowski, Jones, and Orn-
`
`stein (2016), Greenway and Ross (2017), and JAMA (2017)). Second, we treat promotion as
`
`persuasive. In the context