Economics of Pharma Industries

Journal of Economic Literature 2018, 56(2), 397–449
https://doi.org/10.1257/jel.20161327
Economics of the
Pharmaceutical Industry †
Darius N. Lakdawalla*
The pharmaceutical industry accounts for a substantial chunk of the US economy’s

research and development investments, which have resulted in significant medical
breakthroughs. At the same time, the costs of pharmaceutical products continue to
rise, as does pressure to adopt direct or indirect controls on pharmaceutical prices. We
review the economics literature on the pharmaceutical industry, focusing particularly
on its positive and normative implications for the innovation, pricing, and marketing
decisions of pharmaceutical firms. We discuss the major achievements of, and per-
sistent gaps in, the literature, along with lessons for policy. (JEL I11, L11, L65, M31,
M37, O31, O34)
1. Introduction and many other breakthroughs. At the same

time, drug prices continue to rise rapidly in
T he pharmaceutical industry represents

a significant and growing share of
healthcare spending. Innovation in phar-
a variety of markets, especially in the United
States. For all these reasons, the problem of
stimulating affordable pharmaceutical inno-
maceuticals has contributed some of the vation remains a first-order policy challenge.
most important successes in modern med- Many of the most controversial questions
icine—e.g., for the treatment of high cho- about the pharmaceutical industry turn on
lesterol and heart disease, highly active the behavior of pharmaceutical firms. Are
antiretroviral therapy for the treatment of prices for new medicines set too high by
HIV, tyrosine kinase inhibitors (TKIs) for patent-protected firms (Howard et al. 2015)?
the treatment of chronic myeloid leukemia, How does reimbursement of drugs affect the
pace of medical innovation (Kantarjian and
Rajkumar 2015)? Does drug advertising dis-
* University of Southern California and NBER. The
author wishes to acknowledge the National Institute on tort the way physicians prescribe drugs and
Aging for funding (P01 AG033559) and also Talia Bar the way patients use them (Angell 2004)?
for helpful comments on the paper. The author discloses Economists have much to say about the
that he is the Chief Scientific Officer at Precision Health
Economics (PHE), and an investor in its parent company, behavior of pharmaceutical firms, its pre-
Precision Medicine Group. PHE provides consulting and dictors, and its normative implications. The
research services to pharmaceutical and biotechnology purpose of this article is to summarize and
firms.
†
Go to https://doi.org/10.1257/jel.20161327 to visit the critically review the literature on the decision
article page and view author disclosure statement. problems faced by p harmaceutical firms,
397
398 Journal of Economic Literature, Vol. LVI (June 2018)
TABLE 1
Total Employment and Planned R&D Costs Overall and in Selected Industries, 2011
R&D Costs Worldwide R&D
NAICS paid by employees Costs per
Industry code company ($mil) (thousands) employee ($)
Computer and electronic products 334 $77,887 2,951 $26,393

Pharmaceuticals and medicines 3,254 $75,602 1,003 $75,376
Publishing (including software) 511 $39,323 1,185 $33,184
Professional, scientific, and technical services 541 $34,407 2,799 $12,293
Transportation equipment 336 $31,639 2,596 $12,188
Machinery 333 $19,344 1,805 $10,717
All industries 31–33, 42–82 $365,211 29,327 $12,453
Source: National Science Foundation/National Center for Science and Engineering Statistics and US Census
Bureau, Business R&D and Innovation Survey, 2011
with particular emphasis on their implica- publishing, or professional services. New

tions for public policy. drugs must proceed through a lengthy and
Apart from its importance to health-care formalized process of testing for both safety
costs and progress, the pharmaceutical and efficacy. The very public nature of drug
industry is unique and worth studying
development and testing makes corporate
because its institutional features interest secrecy much less useful for protecting phar-
economists (Scherer 2000). First, it is an maceutical R&D investments than for most
research and development (R&D)-intensive other R&D-intensive industries. Thus, phar-
industry. Table 1, above, provides R&D costs maceutical firms rely on patent protection
and employment for the five three-digit and other forms of market exclusivity cre-
North American Industry Classificaiton ated by regulatory fiat.
System (NAICS) industries with the highest While governments regulate market
total R&D costs as of 2011, and compares exclusivity, safety, and efficacy in the phar-
them to the pharmaceutical industry. Total maceutical industry, they also serve as
R&D costs per employee are more than twice a major customer. This creates a rather
as high in pharmaceuticals as in any other unique tension that often, but not always,
industry. Aggregate R&D in the pharmaceu- results in the government exercising mon-
tical industry is second only to the computer opsony power against the patent monopo-
and electronic products industry, which has lies that it created in the first place. Even in
almost three times as many employees. primarily private markets, patent monopo-
As a result of its R&D intensity, today’s lies create strong incentives for customers,
economic policies reverberate over the such as insurance companies, to grow large
long term in the pharmaceutical industry.
enough to exercise countervailing market
Moreover, the pharmaceutical industry faces power. The arrival of similar but not iden-
arguably a much wider array of government tical pharmaceutical innovations from other
regulation and salient public policies than drug manufacturers reinforces this monop-
industries like computers and electronics, sony power, as does the prospect of future
Lakdawalla: Economics of the Pharmaceutical Industry 399
competition by generic manufacturers pro- After testing in a laboratory or on animal

ducing identical drugs. All these sources models—often called “preclinical testing”—a
of competition influence price negotia- manufacturer can submit an investigational
tion with public and private customers, new drug (IND) application to the Food and
and non-price competition like marketing Drug Administration (FDA). If the IND
investments. application is granted, the manufacturer is
To illustrate the economic interplay of all then permitted to proceed to human test-
these peculiar features, we study the three ing. Human testing occurs in three phases,
most important decisions faced by pharma- described later, with progressively larger
ceutical firms: research and development, study populations and more ambitious goals.
pricing, and marketing. Throughout, we If the molecule successfully passes all three
emphasize the various ways in which these phases of testing, the innovator can submit a
decisions interact with each other. The new drug application (NDA), including the
body of this article is divided into four sec- results of all phases of testing. If the FDA
tions. The first provides background on and approves this NDA, the drug can then be
description of the pharmaceutical product marketed. As the product nears its patent
“
life-cycle,” which illustrates the different expiration date, generic manufacturers can
kinds of decision making problems faced by invest in tests designed to show bioequiv-
pharmaceutical firms and their customers. alence of their product with the original
The remaining three sections will then cover branded drug. If successful, they can sub-
the economic analysis of research and devel- mit an abbreviated new drug application
opment, pricing, and marketing, respec- (ANDA). Unlike the NDA, the ANDA need
tively. We conclude with lessons for policy not independently show efficacy and safety.
and future research. It must only show equivalence with the “ref-
erence” product that was already approved.
If the generic firm’s ANDA is approved, they
2. Pharmaceutical Product L
ife Cycle
receive the right to market the generic prod-
The typical pharmaceutical “
life cycle” uct at the expiration (or invalidation) of the
proceeds as follows: original patent.
2.1 Drug Discovery
1. Discovery
2. Development, including testing, The process begins with discovery of a
review, and approval new compound. Drug discovery was once a
3. Oligopolistic competition and p
atent largely serendipitous process. More recently,
extensions however, the process of searching for and
4. Generic competition discovering new molecules has become
much more rigorous and systematic (Scherer
Figure 1, taken from Mossinghoff 2000, p. 1307; Scott Morton and Kyle 2012,
(Mossinghoff 1999), provides a more detailed pp. 772–73). Without delving too deeply into
graphical timeline of this process. After the technical details, s o-called “rational drug
the discovery of a new compound, but well design” begins with a conceptual understand-
before its clinical approval and use, a patent ing of the specific biological targets that must
is filed. The actual patent term depends on be “hit” in order to successfully treat a par-
a variety of factors that we will discuss later, ticular disease. Molecules are then system-
including the length of time spent in FDA atically designed to hit these targets. Put in
review and the success of patent challenges. terms more familiar to economists, rational
Figure 1. Schematic of Pharmaceutical Life Cycle
Source: Mossinghoff 1999. Reprinted with permission of the FDLI.
drug design takes the view that researchers is an important way to protect IP. However,
should begin with a theory of how a drug both the public nature of the drug approval
could address a particular disease, instead process and the low cost of imitation by a
of relying solely on empirical trial and error. competitor that knows the drug’s structure
For example, new cancer agents might work make secrecy an ineffective tool for pro-
by inhibiting substances that facilitate the tecting IP in the pharmaceutical industry
growth of tumors—TKIs for the treatment (Scott Morton and Kyle 2012; Menell and
of chronic myeloid leukemia represent one Scotchmer 2007).
such example.
2.2 Development and Testing
Once a molecule is newly designed, its
inventors typically patent it. While con- A newly discovered, patented molecule first
sidered a dominant, natural, and intuitive undergoes p reclinical, nonhuman testing in
method of intellectual property (IP) protec- laboratory or animal model settings. During
tion within the pharmaceutical industry, pat- this stage of testing, the drug’s toxicity and
ents are typically the least important form of pharmacologic properties are investigated.
IP protection outside it (Cohen, Nelson, and If successful, the process of clinical testing
Walsh 2000). Elsewhere, corporate secrecy begins. On average, it takes a drug around
eight years to move from the start of clinical 2010). As a result, trials must last longer or
testing until market launch (Blume-Kohout enroll more patients in order to record the
and Sood 2013). For illustrative p urposes, same number of patient deaths as in earlier
we outline the steps in the US FDA testing eras of cancer care. This makes it more costly
process. Many regulatory agencies around to generate the same level of statistical power,
the world follow similar processes, although when comparing two alternative treatments.
drug approvals are conducted separately All else equal, continuing progress in med-
by each approval agency. Phase I trials of a ical care will continually expand the cost of
drug are conducted on (often) healthy vol- clinical testing for new medications.
unteers to establish the safety of a drug. In To partially mitigate this tendency, many
the United States, approximately 70 per- innovators employ “surrogate end points”
cent of drugs that enter phase I survive and that are imperfectly correlated with the
move on to phase II, in which alternative true end points of interest, but observed
dosages are tested on sick patients (Adams earlier than death. In cancer, an example is
and Brantner 2006). In addition to establish- “progression-free survival,” which measures
ing dosage, phase II trials provide important the length of time it takes for a patient’s can-
data on safety and efficacy to innovators. For cer to progress to a more advanced stage.
30–40 percent of drugs that began in phase Since progression typically occurs ear-
I, phase II results appear sufficiently prom- lier than death, measuring differences in
ising to move the product onto phase III progression-free survival requires less time
(Adams and Brantner 2006). than measuring overall survival. The use
In phase III, so-called “pivotal” trials are of surrogate end points reduces the cost of
conducted that are designed to establish clinical development, but also increases the
efficacy of the product against one or more uncertainty associated with the results of a
comparators. In phase III, a new drug is trial. This trade-off between development
tested against a placebo or an “active com- cost and uncertainty plays a major role in the
parator”—i.e., an actual medicine, instead choice of trial end point (Stevens et al. 2014,
of a placebo. The choice of comparator in a Lakdawalla et al. 2015). On the one hand, the
phase III trial is fraught with complexity and innovator seeks to contain the costs of clini-
risk for an innovator. Comparison to an active cal testing and development. This argues for
comparator in widespread use heightens the the use of surrogate end points. On the other
risk of a null finding from the trial. However, hand, payers or physicians might be reluctant
comparison to placebo heightens the risk that to accept surrogate end point evidence as
clinicians and payers will fail to recognize the compelling enough to adopt a new therapy.
trial as a reason to abandon the use of com- There is thus a trade-off between speed of
peting medicines that are already in use. development and the probability of market
A similarly complex set of issues surrounds acceptance. This poses problems for patients
the related choices of primary end point and and physicians, who face a similar trade-off
sample size of a phase III trial. The exam- between the speed at which new drugs arrive
ple of cancer is instructive. Most clinicians and the certainty around clinical benefit.
treating patients with cancer focus primarily Once the phase III trial is designed and
on survival as the key clinical end point of conducted, the innovator can choose to sub-
interest. However, advances in the treatment mit a molecule for regulatory approval by
of cancer over the past thirty years mean that a body like the US FDA or the European
many patients with cancer can expect to live Medicines Agency (EMA). Regulators typ-
longer than ever before (Lakdawalla et al. ically evaluate the efficacy established in
clinical trials, weight it against any known et al. 2013). Indeed, the high cost of clinical
safety risks, and incorporate other consid- testing implies the existence of unapproved
erations such as unmet need for new treat- indications with a nontrivial probability of
ments. As an example of the latter, a new clinical efficacy. Simple economics predicts
drug that treats a s o-called “orphan disease” that some drugs should be valuable in indi-
with no effective treatments may be viewed cations that are not formally approved. There
as a higher priority for approval than an may also be scientific reasons for off-label
equally efficacious drug that treats a disease use, particularly if a given molecule happens
with many known, effective treatments. to be effective against a biological target rel-
Among the largest drug companies, roughly evant in multiple disease areas. For exam-
20–30 percent of drugs that began in phase I ple, drugs that enable the immune system to
end up being approved for use (Adams and attack cancer cells are thought to be particu-
Brantner 2006). Approval is given for an larly relevant in the treatment of both met-
“indication,” which is a particular context of astatic melanoma and metastatic renal (i.e.,
use defined by a disease area, a line of therapy, kidney) cancer. This has led to the application
and a patient population. The line of therapy of similar immunotherapies in the treatment
indicates when in the treatment pathway a of both these cancers (Danzon and Ketcham
drug will be used. A “first-line” agent is a 2004; Johnson, Peng, and Sosman 2015).
drug that can be used on a newly diagnosed
2.3 Oligopolistic Competition
patient who has never been treated for his
condition before. A “ second-line” agent Drugs that are approved for use are then
can be used on patients that fail a fi rst-line launched in the marketplace. However, the
therapy, and so on for third-line and later. process of development and testing does not
A single drug can be approved in different end after the launch of a drug. “Phase IV”
indications or lines over time. For example, trials often involve the analysis of the drug
bevacizumab was first approved in 2004 for in real-world settings. These are designed to
use as a first-line or second-line therapy in demonstrate that the drug performs safely
patients with metastatic colorectal cancer.1 and effectively outside the artificial setting
In 2014, on the other hand, it was approved of a clinical trial (Suvarna 2010). In addition,
for the treatment of ovarian cancer, but only some drugs seek approval for additional indi-
in women with recurrence of cancer that cations after they are launched. There may
was previously treated successfully with a be various reasons to seek a new indication.
platinum-based chemotherapy regimen.2
In some cases, it is to boost the usage of the
Once a drug is approved for use, physicians drug within that indication, which may be
may, in principle, prescribe it for any indica- limited by physician reluctance to pursue
tion, even those that lie outside the approved off-label use. In others, new indications can
indication. This is known as “off-label” use, extend patent protection. One example is the
which is quite prevalent in some cases. US law that grants an additional six months
Off-label usage is estimated to be roughly of patent life to a drug that seeks and obtains
30 percent in the treatment of cancer (Conti an indication for use in pediatric patients.
2.4 Generic Competition
1 See http://www.drugs.com/newdrugs/avastin-approved- The final stage in a drug’s life cycle begins
metastatic-colorectal-cancer-21.html.
2 See
with the onset of generic competition that
http://www.drugs.com/newdrugs/fda-approves-
avastin-bevacizumab-plus-chemotherapy-platinum- may accompany the expiration of patent pro-
resistant-recurrent-ovarian-cancer-4109.html. tection. The terms of patents vary and are
impacted by several cross-cutting laws. In methods for modifying genetic material of

general, US patents (except for “design pat- living organisms for the purposes of produc-
ents”) last for twenty years. This underlying ing medicines, agricultural technology, or
patent term, however, has been modified in other products. A biologic drug is a medic-
the pharmaceutical context. The Drug Price inal product of biotechnology. Biologics are
Competition and Patent Term Restoration much more difficult to imitate than con-
Act of 1984 (Public Law 98-417), often ventional small-molecule products, because
referred to as the “ Hatch–Waxman Act,” their production depends on the cultivation
recognized that part of a drug’s patent life of a specific biological “cell line” that may
occurs prior to the date of FDA approval. be difficult to reproduce. Thus, even if the
Since the innovator cannot recover prof- structure of a biologic molecule is known,
its for this p relaunch portion of the patent it may be difficult or even impossible to
term, the Hatch–Waxman Act extended the replicate an innovator’s product without
term of pharmaceutical patents by up to physically possessing the cell line used to
five years, on top of the baseline period of produce it. The Hatch–Waxman Act does
patent protection. not provide a pathway for generic compe-
The H atch–Waxman Act also establishes tition to biologic products. More recently,
the process by which generic manufactur- however, legislative changes in the United
ers can seek approval from the US FDA for States and Europe have introduced the con-
generic small-molecule drugs. The act estab- cept of “biosimilars.” For example, the US
lishes the ANDA process, which allows the Affordable Care Act defines a biosimilar as
generic applicant to rely on the efficacy and having “no clinically meaningful differences
safety data generated by the original innova- between the biological product and the ref-
tor. Moreover, the Hatch–Waxman Act also erence product in terms of safety, purity,
provides incentives for generic manufactur- and potency” (Blackstone and Fuhr 2013).
ers to challenge patents in court, under its Because biologic products are produced in
“Paragraph IV.” Patent challenges entail a living systems, their structure is more vari-
free-riding problem similar to that of inno- able than that of small-molecule products,
vation itself: it is costly to invalidate a patent which have a single, well-defined chemical
in court, yet a whole host of firms could ben- structure. This variability becomes signifi-
efit from a successful challenge. The result is cant because the cell lines or other organic
that no firm has adequate incentive to chal- systems used to produce biologic medicines
lenge a patent. Paragraph IV of the H atch– are not generally in the public domain. This
Waxman Act provides generic manufacturers asymmetry of information provides another
with six months of generic exclusivity in the means of protecting IP.
event that they successfully challenge the From an economic point of view, produc-
patent of a branded manufacturer. During ing a biosimilar involves greater barriers to
this period, only two firms can produce the entry than producing generic s mall-molecule
molecule (Mossinghoff 1999, Sanjuan 2006). drugs, because a new cell line or production
process must be developed. Variable costs
2.5 Biologic Drugs
are also higher, although recent advances
The life cycle above applies most fully with the use of plants—rather than ani-
to “ small-molecule” or conventional mals—to produce biologic products hold the
pharmaceutical products. Biologic, or promise of substantial cost reductions in the
“
large-molecule,” products represent a future. Finally, there is less certainty that a
different case. Biotechnology comprises biosimilar will possess identical safety and
efficacy properties as the reference molecule one for investments into discovery, and one
(Blackstone and Fuhr 2013). For these and for the outcome of this investment to be real-
related reasons, many economists believe the ized. The critical assumption is the monotonic
competitive effects of biosimilars will be less relationship between innovative investments
pronounced than the competitive effects of and the probability of successful discovery.
small-molecule generic drugs. While generic Define Ias the total level of investment
drugs are regarded by the marketplace as into innovative activity, and p(I)as the
near-perfect substitutes, most of the early probability of a successful discovery, where
experience with biosimilars suggests they are p′(I)  > 0and p ′′(I)  < 0. Define D as the
regarded as therapeutic competitors, much state of the world in which a discovery occurs,
the way a new branded drug within the same and Nas the state in which no discovery
class would be viewed alongside another occurs. E(π | D)is the innovator’s expected
branded competitor. For example, a report profit, conditional on discovery, and similarly
from the Federal Trade Commission (FTC) for E(π | N). Innovation investments will be
reaches this conclusion based on its findings undertaken if and only if E( π|D)  > E(π | N).
that biosimilars face higher entry costs than To allow for the possibility that the firm
generic small-molecule drugs, and that bio- may not have to bear all the cost of inno-
similars will have a harder time gaining mar- vation—e.g., if some portion of research
ket acceptance than small-molecule generics is publicly funded—define ϕ (I)  ≤ Ias the
due to the possibility for differences in safety firm’s private cost of investment, I. Finally,
and efficacy (FTC 2009). define r as the cost of capital.
The resources that the firm expects in
the event of discovery equal E ( π|D)  −
3. Research and Development Investments
(1 + r)ϕ(I), and in the event of no discovery,
We now turn to the economics of R&D E(π | N)  − (1 + r)ϕ(I). Therefore, the pri-
investments by pharmaceutical firms. To vately optimal level of innovation is given by
organize the discussion, we present a sim- the solution to:
ple theory of innovation investments, due to
Nordhaus, that lies at the heart of modern    p(I)E( π|D)  + (1 − p(I))E(π|N) 
max
I
theory (Nordhaus 1969). We then discuss
the empirical evidence on the supply- and − (1 + r)ϕ(I).
demand-side determinants of innovation

that the theory identifies and predicts. This has the fi
rst-order condition:
Finally, several mainstream critiques of the
“simple theory” will be presented and dis- p′(I)[E(π|D)  − E(π|N)]  = (1 + r ) ϕ′(I).

cussed, along with their empirical and nor-
mative implications. This condition implies that the marginal
private cost of capital (right-hand side) equals
3.1 The Simple Positive Theory of
the marginal private return of innovation
Innovation Investments
investment (left-hand side). This simple and
In his seminal paper on incentives for inno- intuitive condition has a number of import-
vation, Nordhaus argues that investments ant implications. First, all else equal, innova-
into the discovery of innovation rise with the tion investment rises with the expected gain
profits expected from successful discovery from discovery, which is the term in square
(Nordhaus 1969). This point can be made brackets. Second, investment also rises with
most simply using a model with two periods— increases in the marginal productivity of
investment, p′(I)
. Third, increases in the in expected market size stimulate innovation
marginal cost of capital depress innovation effort. There is widespread empirical sup-
investments. We discuss the empirical evi- port for this relationship, but much debate
dence on each of these implications in turn. over its magnitude. Acemoglu and Linn
(2004) is arguably the first study to offer a
3.2 Empirical Evidence on the Returns to
causal estimate. They find that a 1 percent
Innovation
increase in expected market size is associated
The simple theory predicts that increases with a 4 percent to 6 percent increase in the
in expected p (π | D),
ost-discovery profits, E number of new molecular entities entering
stimulate innovation. This implies that the market.
growth in consumer demand for pharma- Acemoglu and Linn (2004) exploit plau-
ceuticals stimulates innovation investments. sibly exogenous variation in market size for
“Consumer demand” encompasses pref- different drug classes driven by demographic
erences, prevalence and severity of illness, trends in the US population. They observe
and a range of d emand-side public policies that different drug classes are used by patients
(e.g., subsidies, price regulations, or manda- at different ages—e.g., contraceptives by
tory utilization). Furthermore, it implies that younger groups, and hypertension medica-
stronger market exclusivity stimulates inno- tion by older groups. Thus, relative growth
vation investment by boosting post-discovery in old or young populations, driven by demo-
profits. Market exclusivity protection within graphic trends, drives demand for some drug
the pharmaceutical industry encompasses classes but not others. If innovation rises in
patent protection, regulatory exclusiv- drug classes with market expansion, but not
ity (e.g., data exclusivity), and restrictions in classes without it, this provides support for
on pathways for the approval and entry of the underlying theory. Relying on this idea,
generic competitors. Acemoglu and Linn find that innovation
responds to expected market size as theory
3.2.1 Consumer Demand
predicts. They find that a 1 percent increase
Expansions in market size boost pharma- in expected market size is associated with a
ceutical profits and are thus predicted to 4 percent to 6 percent increase in the num-
stimulate innovation. On the other hand, ber of new molecular entities entering the
price ceilings or related regulations constrain market. These results are robust to controls
profits and do the opposite. Even though for a variety of supply-side factors such as
economic theory seems clear and intuitive public R&D funding and drug class-specific
on these points, persuasive empirical evi- time trends.
dence remains valuable, because the effect However, several authors contend that
of product demand on innovation still gen- Acemoglu and Linn (2004) overstate the
erates substantial controversy among physi- true causal parameter linking market size
cians, public health researchers, and policy to innovation (de Mouzon et al. 2011).
makers (Angell 2004). This may occur, for example, if increased
innovation in one drug class comes at the
3.2.1.1 Market Size and Innovation
expense of reduced innovation in other
We begin by reviewing the literature estab- drug classes. Suppose, for instance, that the
lishing the causal link between expected future demand for oncology products rises
market size—often measured as aggregate as a result of population aging, but that the
consumer w illingness to pay—and innova- demand for childhood vaccines remains con-
tion effort. Theory implies that expansions stant. If innovators face imperfect capital
markets, they may finance the increase in these results with the innovation elasticity
oncology innovation by cutting spending estimates from Acemoglu and Linn would
on childhood vaccines. Ignoring the latter imply an increase of between 49 percent
effect would overstate the true relationship and 83 percent in new drug development for
between market size and total innovation. the average drug class. The B lume-Kohout
While debate persists over the size of the and Sood estimates are generally within this
innovation-market size elasticity, few dis-
range, if toward the more conservative end
agree that growth in expected market size in many cases.
fuels more innovation. Indeed, a large lit- These studies focus on the quantity of
erature demonstrates the effects of various innovation that derives from expansions
demand-side policy changes on the incentives in market size. Recently, the literature on
to innovate. To take one example, direct sub- market size has begun to question the qual-
sidies for the purchase of vaccines have stim- ity of innovation that results from at least
ulated R&D in vaccine markets (Finkelstein some types of market expansions. Dranove,
2004). To take another, public subsidies for Garthwaite, and Hermosilla (2014) argue
prescription drug insurance in the United that Medicare Part D indeed spurred innova-
States—specifically, the Medicare Part D tion into diseases that target the elderly, but
program—have stimulated investment in the resulting drugs were more likely to be in
the discovery of drugs targeted toward the diseases with multiple existing treatments,
program’s beneficiaries (Blume-Kohout and rather than diseases with few available ther-
Sood 2013). Specifically, Blume-Kohout and apies. For example, the tenth b eta blocker
Sood assess how Medicare Part D affected medication to treat high blood pressure and
pharmaceutical research and development. cholesterol might provide little if any incre-
They measure R&D using data on: (i) the mental social value, even if it is technically a
number of drugs in clinical development different molecule than the other nine drugs
by therapeutic class, and (ii) research and in its class. This finding, however, suffers
development expenditures by firms. They from problems of interpretation, given the
demonstrate that the passage of Medicare argument made by others that “follow-on”
Part D coincided with significantly higher therapies are not necessarily viewed by con-
pharmaceutical research and development sumers as highly substitutable with the pio-
for drug classes with higher Medicare mar- neer therapies that they follow (Jena et al.
ket share, and for firms specializing in 2009). There are a number of possible rea-
higher-Medicare-share drugs. sons why this might be true: follow-on ther-
The B lume-Kohout and Sood (2013) apies might entail fewer side effects, serve
paper supports the magnitudes estimated by additional patient populations that were
Acemoglu and Linn (2004). The logic pro- unresponsive to the pioneer agent, or exhibit
ceeds as follows. A study by Duggan and Scott greater efficacy. Cholesterol-lowering statins
Morton (2010) concludes that Medicare Part are a good example, where the pioneer prod-
D increased pharmaceutical revenues by uct (simvastatin) exhibits substantially lower
roughly 33 percent for drugs with Medicare efficacy than subsequent products (e.g., ator-
market share of 100 percent. For a drug vastatin) (Grabowski et al. 2012). If some
class with average Medicare market share or all of these points prove true, follow-on
in their sample (42 percent), Duggan and therapies may generate significant value to
Scott Morton’s result translates to a 14 per- consumers.
cent change in pharmaceutical revenues Nonetheless, while there are differences
following Medicare Part D. Combining of opinion regarding magnitudes, there is
strong consensus that growth in pharmaceu- Department of Commerce reviewed pricing

tical demand stimulates innovation. A notable in eleven OECD countries and found that,
exception presents itself in emerging mar- for patented drugs that were best sellers
kets. Kyle and McGahan (2012) have stud- in the United States (one of the least regu-
ied the T
rade-Related Aspects of Intellectual lated markets), the prices in other OECD
Property Rights (TRIPS) agreement, which countries were 18 to 67 percent less than
mandated minimum standards for patent US prices. They conclude that price dereg-
enforcement among members of the World ulation across these eleven countries would
Trade Organization (WTO). Among emerg- increase pharmaceutical revenues by 25 to
ing-market members of the WTO that did 38 percent (US Department of Commerce
not previously enforce patents, TRIPS trig- 2004). Similarly, Ekelund and Persson
gered patent protection. Kyle and McGahan (2003) find that in contrast to the less reg-
show that the introduction of patent protec- ulated US market, prices of new drugs in all
tion in emerging markets had no detectable classes fall faster in the regulated market of
effect on global pharmaceutical innovation Sweden. Another study, Martikainen, Kivi,
for diseases prevalent in these markets. and Linnosmaa (2005), finds that whole-
However, they find—consistent with the rest sale prices for newly launched reimbursable
of the literature—that patent protection in pharmaceuticals were highest in countries
developed countries did stimulate innova- where manufacturers are free to set their own
tion. These findings do not necessarily indict prices. Finally, Danzon and Chao (2000) find
the Nordhaus theory of innovation. For a negative relationship between strict price
example, it may be that patent protection in regulation and the price of widely approved
poorer countries offers insufficient rewards molecules. While the underlying theoreti-
for innovators to overcome the fixed costs of cal hypothesis is intuitive, these studies are
embarking on research programs into trop- limited by their reliance on cross-sectional
ical diseases. However, Kyle and McGahan variation in revenues or prices. They remain
(2012) do strike a cautionary note about vulnerable to heterogeneity across coun-
universal implementation of strong patent tries in type of regulation and other price
protection. determinants.
There are some studies that address the
3.2.1.2 Price Regulation and Innovation
heterogeneity problem by analyzing lon-
While patent protection guards prof- gitudinal data and comparing pharmaceu-
its, price regulation reduces profits. Since tical expenditure before and after policies
profitability appears to influence innovation take effect (Pavcnik 2002, Pekurinen and
empirically, this would seem to imply that Häkkinen 2005). For example, Pavcnik
price regulation depresses innovation, too. (2002) estimated a 10 percent to 26 percent
The empirical literature testing this theoret- decrease in drug prices as a result of a refer-
ical link consists of two kinds of studies. The ence-pricing policy introduced in Germany
first establishes the extent to which price after 1989. Similarly, Brekke, Grasdal, and
regulations reduce revenues, and whether Holmas (2009) estimated that introduction
or not innovative products are dispropor- of reference pricing in Norway reduced
tionately affected. The second quantifies the both b rand-name and generic prices within
reduced-form effect of price regulation on the reference group, with the effect being
innovative activity. stronger for brand names. Pekurinen and
Price regulations appear correlated with Häkkinen (2005) suggested that voluntary
lower prices. For example, a study by the US generic substitution and prescribing policies
had no effect on expenditures in Finland, 3.2.2.1 Background on Alternative Forms

but that compulsory generic substitution of Market Exclusivity Protection
decreased prices and led to cost savings in
the first year after introduction. Sood et al. While most people think of patents as the
(2009) exploit variation in a variety of phar- only source of IP protection, other mecha-
maceutical regulations in nineteen OECD nisms are becoming gradually more import-
countries over the period of 1992 to 2004. ant. There are at least three major kinds of
They find that most regulations reduce phar- protections that provide significant advan-
maceutical revenues significantly and that tages to pharmaceutical innovators:
the cost-reducing effects of price controls
accumulate with time since implementa- 1. Patent exclusivity
tion. Moreover, all or most of these stud- 2. Regulatory exclusivity
ies might underestimate the true effects of 3. Manufacturing challenges that serve
regulations if countries implement regula- as barriers to entry for large-molecule
tions in response to rising pharmaceutical products
spending.
Although the evidence on the effects of As discussed earlier, patents remain in force
price regulation on pharmaceutical reve- for twenty years, absent any extensions or
nues is fairly substantial and convincing, the challenges, and the Hatch–Waxman Act
evidence on the effects of price regulation typically extends this term by an additional
on pharmaceutical innovation is not as well five years, to mitigate time lost in the FDA
developed. Vernon (2005) examines whether approval process. Even though the approval
firms with a larger proportion of sales in process consumes part of the patent period,
the US market invest more in pharmaceu- patents nonetheless possess the advantage of
tical R&D. The argument is that firms with long duration. The disadvantage, however,
a larger US sales share are less exposed to is that patents can be challenged in court
price regulation, as most other major phar- and invalidated, as the Hatch–Waxman Act
maceutical markets have some sort of price explicitly allows for and encourages.
regulation in place. The empirical evidence In contrast to patents, regulatory exclu-
is consistent with the hypothesis that price sivity eliminates the possibility of judicial
regulation reduces pharmaceutical R&D. challenges, but shortens the period of pro-
However, it is also possible that firms that are tection. One form of regulatory exclusivity is
not focused on innovation tend to operate in established by the Orphan Drug Act (ODA),
non-US markets. which grants an innovator an exclusive
license to market an eligible orphan drug for
3.2.2 Market Exclusivity Protection
a specific indication for seven years after its
Above and beyond price regulation, pol- approval for marketing (Thamer, Brennan,
icies that protect or fail to protect IP and and Semansky 1998). Passed in 1983, the
market exclusivity might also influence US ODA defined an orphan drug as one
innovation investments. IP protection in the that treats a disorder affecting fewer than
pharmaceutical industry is often viewed as 200,000 people in the United States. Orphan
more critical than in other R&D-intensive drug exclusivity is the strongest form of reg-
industries because of the high costs of ulatory exclusivity. It cannot be challenged in
development combined with the low costs court, and no competitor can gain approval
of imitation (Scott Morton and Kyle 2012, for a drug treating the relevant orphan
p. 772). indication.
A more widely applicable form of regu- data exclusivity. Clearly, longer patents will
latory exclusivity is “data exclusivity,” which raise E(π | D)and thus stimulate innovation
bars generic manufacturers from using the incentives. There is not a large empirical
safety and efficacy data generated by a pio- literature directly testing this implication,
neer firm for some period of time (Goldman perhaps because it seems obvious. However,
et al. 2011). Data exclusivity poses a substan- there are some studies that indirectly vali-
tial barrier to entry due to the high cost of date it. For example, in order to comply with
conducting clinical trials to generate safety the provisions of the TRIPS agreement, the
and efficacy data. The H atch–Waxman Act United States had to extend patent length.
provides pioneer firms with five initial years Research on overall patent counts, rather
of data exclusivity, plus an additional three than pharmaceutical patents alone, suggests
years for supplemental applications. In addi- that this extension in patent length stimu-
tion, the Food and Drug Administration lated innovation (Abrams 2009).
Modernization Act of 1997 (FDAMA) pro- A more complex issue is the possibility
vides a six-month extension for pediatric of “endogenous” patent length. The value
applications. By comparison, the duration of of market exclusivity creates incentives for
data exclusivity in Europe is ten years plus an firms to obtain patents and lengthen them
additional year if a new indication is added by any available means. This includes the
that provides significant clinical benefits over filing of multiple patents with staggered
existing therapies. expiration dates, reformulations designed to
Finally, as discussed earlier, there are nat- prolong the availability of market exclusivity,
ural barriers to entry in large-molecule bio- or s o-called “reverse payments” designed to
logic products. In this segment of the market, delay generic entry (Scott Morton and Kyle
production processes are much harder to 2012, pp. 8 00–801). “Reverse payments” are
mimic and can be protected under corporate settlements paid by branded manufacturers
secrecy laws (Menell and Scotchmer 2007), to generic manufacturers in the context of
rather than by variable patent or regulatory patent litigation. The reverse payments are
exclusivity regimes. typically coupled with an a greed-upon date
of market entry for the generic firm. Prior
3.2.2.2 Design of IP Protection and
research suggests that the announcement of
Incentives to Innovate
reverse payments appears to coincide with
Patents themselves vary in the extent increases in the stock prices of the branded
of protection they afford. Patent length is drug companies making the payment
the amount of time for which an innovator (McGuire et al. 2016). This suggests that
receives market exclusivity. Patent breadth reverse payments may be linked with higher
governs the scope of market exclusivity, expected profits for the branded manufac-
and the scope of product variations across turer, presumably because it delays generic
which competitors cannot infringe. Both entry past the date that would have been
patent length and scope are subject to legal expected without the payment.
challenges and interpretation, which create
uncertainty. Patent Breadth.—Economists often view
valid patents as perfect bulwarks against imi-
Patent Length.—Innovation incentives tation, but the reality is more complicated.
depend on the length of market exclusiv- As discussed earlier, patents can be fully
ity, which is determined by the interaction invalidated, but they can also be narrowed
between patent length and the length of or expanded in scope. Two legal doctrines
g overn changes in patent breadth: the “doc- against challenges to patent scope or length.
trine of equivalents,” and “the reverse doc- We are not aware of any empirical literature
trine of equivalents.” testing this possibility directly, however.
In principle, competitors could skirt pat-
ent protection by making minor changes Empirical Evidence on the Value of the
to protected products. Recognizing this Patent Mechanism.—From a normative
loophole, courts have asserted the doctrine standpoint, patents possess several import-
of equivalents, which protects an inventor ant virtues. First, patents create a natural
against competing products that are “insub- scheme for P areto-improving innovation.
stantially different” from the protected Innovators sell their inventions to consum-
product. On the other hand, the reverse ers only if the consumer’s w illingness to pay
doctrine of equivalents limits the doctrine exceeds the price. Second, patents decentral-
of equivalents by arguing that new products ize innovation and avoid the agency or mor-
performing the same function better than al-hazard costs associated with p ublic-sector
an already-patented product do not infringe control over research. Decentralization also
upon the patent. allows well-informed inventors to make deci-
In the pharmaceutical context, the doc- sions about how to allocate research dollars.
trine of equivalents revolves around whether However, the empirical role of patents
a biochemical modification constitutes an in driving innovation remains controver-
“insubstantial” change or not. Substantiality sial due to several concerns about the sim-
is often judged according to the “triple iden- ple theory of innovation. First, the simple
tity test,” which identifies patents’ similarity theory assumes that new inventions stand
by whether they perform the same function alone and do not depend on previous dis-
to a varying degree, perform that function in coveries. However, if innovation is cumu-
basically the same way, and yield the same or lative in the sense that new inventions rely
similar results (Albainy-Jenei 2006). critically on prior knowledge, the effects of
From an economic point of view, uncer- patents on innovation become more prob-
tainty in patent scope, as opposed to length, lematic (Boldrin and Levine 2008; Encaoua,
creates uncertainty in the level of per period Guellec, and Martinez 2006). Boldrin and
monopoly profits. In the canonical model, Levine argue that innovators can enjoy com-
the net present value of expected profits is petitive rents from innovation even in the
linear in per period profits, implying risk absence of patents, and these rents might be
neutrality in patent scope. Firms would thus enough to promote innovation. While their
be indifferent to an increase in the riskiness insight might be applicable in some contexts,
of profits, so long as expected profits remain its relevance to the pharmaceutical market
unchanged. is less clear. Small-molecule pharmaceutical
The situation may become more compli- products are quite easily imitated, particu-
cated if some imperfection introduces risk larly since approval requires full disclosure
aversion over profits. For example, if internal of a drug’s composition, safety, and efficacy.
funds are cheaper than external funds, this Nonetheless, even though patents likely
increases the value of smooth cash flow over stimulate pharmaceutical innovation, empir-
time and penalizes the firm with a variable ical evidence suggests they might have
flow of profits. Later, we discuss the literature some effects that reduce dynamic welfare.
that suggests this possibility. Firms facing Williams (2010) studies the impact of pat-
capital market imperfections may thus invest ents granted to the private firm Celera, for
more resources in protecting themselves gene sequencing data. Celera obtained IP
over data from a gene that it sequenced, but to innovators and thus the rate of innova-
lost it if the publicly funded Human Genome tion. If innovators capture a higher share
Project re-sequenced that gene. As a result, of social surplus, this distributive effect can
sequencing data discovered first by Celera drive innovation higher and thus produce
were costly to access, while all sequencing “real”—rather than purely distributive—
data from the Human Genome Project were consequences. For example, the share of
free. Williams finds that the IP-protected real cost reductions captured by innovators,
genes exhibited 30 percent less follow-on while purely distributive in a static sense,
innovation than genes that were sequenced also increase real investments in innovation.
but not protected. While this point is not If innovators capture “too high” a share of
directly relevant to pharmaceutical patents these cost-reductions, the result is ineffi-
per se, it has important implications for the ciently high innovation, and vice versa. In
patent protection of genetic data, which this way, the theory of innovation ties static
influences the arrival of new treatments. distributional issues to dynamic efficiency.
More directly relevant to pharmaceu-
3.3 Empirical Evidence on the Cost of
ticals, Budish, Roin, and Williams (2015)
Innovation Investment
find that fixed patent length theoretically
and empirically discourages research with Pharmaceutical innovation is costly.
long-term payoffs. It takes longer to prove Estimates suggest that the average capital-
the efficacy of a new drug when patients ized cost of bringing a new biopharmaceuti-
exhibit longer baseline survival. For exam- cal to market is $1.2 billion in total (DiMasi
ple, if it takes ten years for any patients to and Grabowski 2007). More recent estimates
die of a disease, it will take at least ten years suggest this number may exceed $2.5 bil-
to show a survival benefit. Budish, Roin, and lion,3 although this number has been met
Williams use this insight to demonstrate that with skepticism from some quarters. Indeed,
research investments are lower in cancer estimates based on publicly available data
types where patient survival is longer, and have suggested lower numbers (Adams and
vice versa. Interestingly, they note that this Brantner 2006).
correlation disappears when innovators can While controversy continues to swirl
use surrogate end points—like time until around the level of cost associated with bio-
disease-progression—that can be observed pharmaceutical innovation, the more salient
well before a patient dies. The bias toward issue, from an economic perspective, is how
innovations with s hort-term clinical benefits changes in the cost of innovation affect the
can be thought of as a distortion introduced behavior of firms. In this section, we will
by fixed patent length. review the empirical evidence on the effec-
Pauly (2009) notes another potential dis- tiveness of “push,” or supply-side mecha-
tortion created by patents. Pharmaceuticals nisms for encouraging innovation. All such
that reduce real resource costs—e.g., if a mechanisms promote R&D investment by
new drug keeps people out of hospitals and lowering the cost of innovation. We focus on
doctors’ offices—might not reduce con- two strands of literature. The first estimates
sumer spending, because the c ost-reduction the effects of policy changes that reduced the
might be partially or wholly captured by the cost of innovation. The second determines
innovator in the form of a higher price. More how cash flow or current period revenues
generally, the “distributive” consequences of
patents might end up having efficiency con- 3 See http://csdd.tufts.edu/files/uploads/cost_study_
sequences, because they affect the returns backgrounder.pdf.
affect R&D spending by influencing the cost (2003) analyzed the effects of the ODA on
of capital. pharmaceutical innovation and found that
it significantly increased new drug intro-
3.3.1 Effects of Pharmaceutical Research
ductions. It is not wholly straightforward to
and Development Costs on Innovation
interpret the effects of the ODA, because it
A variety of studies estimate the effects of blended both supply-side and d emand-side
public policies aimed at reducing the costs incentives. On the supply side, the ODA
of R&D. Most find that pharmaceutical firms offered a 50 percent tax credit on clinical tri-
significantly increase R&D expenditures als for orphan drugs, clinical research grants,
and innovation in response to such policies. and FDA advice and counseling. On the
Examples include: tax credits (McCutchen demand side, the ODA also included a guar-
1993) and tax deductions (Mansfield 1986) anteed s even-year market exclusivity period.
for research and development expendi- The FDA has characterized this exclusivity
tures; tax credits for clinical trial expendi- period as the most important. Regardless,
tures (Lichtenberg and Waldfogel 2003, Yin while the net effect of the law may blend
2008); faster drug review and approval pro- these various supply-side and d emand-side
cesses (Ridley, Grabowski, and Moe 2006; incentives, Lichtenberg and Waldfogel find,
Berndt, Gottschalk, and Philipson 2005); on balance, that in the period (beginning
and direct public expenditures on basic sci- in 1979) before the implementation of the
entific research (Toole 2007, Blume-Kohout ODA in 1983, the number of orphan drugs
2012, and Ward and Dranove 1995). increased at about the same rate as the num-
Public policies that lower the private cost ber of other drugs. By 1998, there were
of innovation investments, ϕ (I)
, ought to more than five times as many orphan drugs
stimulate both spending and drug discov- as there had been in 1979, and fewer than
ery. Assessing the effect on drug discovery twice as many non-orphan drugs.
is important, because it reveals whether or In a subsequent study, Yin (2008) also
not the additional public spending is being finds that the ODA had a significant impact
directed in a productive manner that actually on rare disease drug development, both in
results in the development of new drugs. terms of new clinical trials for established
McCutchen (1993) studies the impact of rare diseases and by the recasting of existing
R&D tax credits on pharmaceutical R&D drugs as treatments for rare diseases. The
expenditures. The tax credit appears to latter effect is not a direct test of the sim-
coincide with an increase in pharmaceuti- ple innovation investment theory because
cal R&D spending. Quantitatively, each $1 it does not involve increases in investment
of tax credit led to an additional 29.3 cents that produce new drug discoveries. Rather,
of R&D spending. The McCutchen study it is more akin to rent-seeking behavior by
has been buttressed by other work, as well. pharmaceutical firms seeking to gain market
For example, Mansfield (1986) finds that tax exclusivity for a drug by repurposing it. A
deductions in the United States, Sweden, and related rent-seeking phenomenon is known
Canada increased R&D by about 1–2 percent colloquially as “salami slicing,” wherein a
in these countries. The ratio of tax-induced manufacturer seeks to win multiple rare
R&D to foregone government revenue indications for a single drug (Kanavos and
ranged from 30 percent to 40 percent. Nicod 2012). As a result of salami slicing,
Others have found that reductions in the an orphan drug may end up treating a fair
cost of investment have stimulated drug number of patients in the aggregate. Its
discovery as well. Lichtenberg and Waldfogel orphan status becomes more of a regulatory
than an economic designation. The complex R&D expenditures. Each $1.00 increase in
nature of the ODA and the apparent arrival expenditures on basic and clinical research
of rent-seeking behavior make it a less direct correlates with $8.38 and $2.35 more phar-
test of how innovation costs affect spending maceutical R&D expenditures, respectively.
and discovery. In a subsequent paper, B lume-Kohout
An alternative policy example is the (2012) finds that public expenditures on basic
Prescription Drug User Fee Act (PDUFA) of research increase the number of clinical tri-
1992, and the subsequent PDUFA of 1997. als initiated by the pharmaceutical indus-
These acts were designed to reduce the time try. Earlier research by Ward and Dranove
it takes for the FDA to review and approve (1995) estimates that a one percent increase
new drugs. Once again, PDUFA blends both in government-funded basic research in a
demand-side and supply-side factors. On the particular therapeutic category causes a 0.76
demand side, PDUFA increases the expected percent increase in pharmaceutical industry
net present value of revenues by accelerating R&D within that category.
the launch of new drugs. On the supply side, On balance, the literature suggests that
it reduces the length of time during which public policies designed to reduce the cost
an innovator must interact with the FDA. of innovation investment stimulate R&D
Berndt, Gottschalk, and Philipson (2005) spending, as predicted by economic theory.
study the impact of the PDUFA of 1992 Less clear, however, is whether this results
and 1997. They find evidence that PDUFA in more drug approvals, although it is diffi-
reduced approval times beyond what cult to conceive of a circumstance in which
would have been observed without these higher spending by p rofit-maximizing firms
acts (6–7 percent annual declines during should consistently have no effect on drug
PDUFA I and 3–4 percent during PDUFA discovery.
II). However, they also find that the cumu-
3.3.2 Current-Period Revenues and the
lative number of new molecular entities
Supply of Innovation Investments
following the passage of PDUFA remained
largely unchanged. Thus, their findings sug- A less obvious determinant of innovation
gest that the reduction in approval times had spending is the cost of, and access to, capi-
little effect on R&D investments. One lim- tal enjoyed by the pharmaceutical industry.
itation of this study is the reliance on a treat- Interestingly, cash flow appears to be one of
ment effect that is uniform across the drugs the most empirically important determinants
studied. As a result, they relied solely on of private R&D spending by private pharma-
time-series variation and lacked a contempo- ceutical firms (Grabowski 1968, Grabowski
raneous control group. This makes it impos- and Vernon 2000). This may be evidence
sible to separate the effects of PDUFA from of capital market imperfections in the
other secular trends in innovation. Much like pharmaceutical industry—e.g., some have
the analysis of ODA, PDUFA is not a clean hypothesized that the internal cost of capi-
test of reductions in the cost of innovation tal is lower than the cost of external capital
investment. raised through the issuance of debt or equity
Public expenditures on basic research (Hubbard 1998). Indeed, within the phar-
and training of a scientific workforce might maceutical industry, s hort-term deviations in
also reduce the costs of developing new profitability appear to predict R&D expendi-
drugs. Toole (2007) estimates that public tures (Scherer 2001). A purely neoclassical
expenditures on basic and clinical research firm facing a perfect capital market would
correlate positively with pharmaceutical base its investment decisions on expected
return alone, not how much cash it has on out of internal funds before driving the rate
hand. of return on investment below r   0, it may then
However, in a simple model with capital turn to external sources of finance, but only _
market imperfections, firms will exhaust if the marginal return on investing the    I  th
their supply of internal capital before seek- dollar exceeds the cost of external capital.
ing external funding. In this environment, Second, if the firm is c apital constrained,
short-run bursts in profitability may tempo- the marginal return on investment _ will vary
rarily lower the rate of return on investment with the level of internal capital,    I  . Increases
within the firm. A simple extension to the in the availability of internal funds will boost
canonical theory of innovation investment investment and decrease the marginal return
illustrates this point. Define I  0 and I  1 as an demanded on capital investment. This fea-
investment from two different sources of ture can be used to distinguish capital con-
capital, with costs r  0 and r  1, respectively, straints models from expectations-based
and r    < r  . Typically, we assume that inter-
0 1
explanations. If, for example, firms use tem-
nally generated capital—e.g., from sales—is porary bursts in profitability to infer long-run
cheaper than capital that is obtained through increases in profitability, then short-term
debt or equity financing. In this case, there- increases in profits will be associated with
fore, we would think of I 0 as investment higher return on investment. The capital con-
from internally generated capital and I  1 as straints model, however, implies the opposite
external capital. We can then write: result, which appears to be consistent with
the empirical literature studying the phar-
m0 ax1   p(I  0  + I  1)E( π|D)  maceutical industry (Bhagat and Welch 1995;
I   ,I  
Giaccotto, Santerre, and Vernon 2005).
+ (1 − p(I  0  + I  1))E( π|N)  The empirical literature appears to sup-
port the capital constraints theory. A num-
− (1 + r  0)ϕ( I  0)  − (1 + r  1)ϕ( I  1) ber of papers find that cash flow is the most
_ important private determinant of R&D
s.t. I  0  ≤    
I  spending. Grabowski (1968) and Grabowski
and Vernon (2000) study the determinants of
I  1 ≥ 0. R&D expenditures of major pharmaceuticals
_ firms during the periods 1959 to 1962, and
Here,     I  represents the supply of cheaper 1974 to 1994, respectively. They find that
internal capital. Associating the multipliers cash flows predict R&D as a fraction of sales.
λ  0 and λ  1 with these constraints, respec- The normalization with respect to sales helps
tively, this has the first-order conditions: adjust for differences in size across firms,
although it does introduce the possibility
p′(I)[E(π|D)  − E(π|N)]  = (1 + r  0) ϕ′(I  0)  + λ  0 of mechanical bias induced by the reverse
causality running from sales to cash flow.
p′(I)[E(π|D)  − E(π|N)]  = (1 + r  1) ϕ′(I  1)  − λ  1. Nonetheless, they find that cash flow is even
more quantitatively important than pharma-
This slightly expanded model generates ceutical profit expectations as a determinant
the following implications. First, the firm of the R&D to sales ratio. This relates to the
takes a sequential approach when financing more general point, echoed by others, that
its investments. It will exhaust the cheaper the source of financing will affect investment
(internal) source of funds first, before relying behavior in the existence of capital market
on external financing. In the event that it runs imperfections (Hubbard 1998).
Similarly, Scherer (2001) finds that devia- of areas, although some doubts might arise
tions from trends in gross profit margins and for large, established firms with seemingly
R&D expenditures are highly correlated. high-quality access to capital markets.
Under the simple model without any capital
3.4 The Determinants of Innovation
market imperfections, there would be no link
Productivity
between these variables because short-run
fluctuations in profitability should not impact We next discuss literature on the deter-
the investment behavior of a forward-looking minants of innovation productivity, which
firm. However, if internal capital is cheaper, depends on both scientific and regulatory
the s hort-run availability of profits may influ- considerations. From a scientific standpoint,
ence R&D expenditures. various economic, clinical, and other factors
Several empirical studies have found that influence the probability that a new drug
current period cash flow predicts R&D. discovery ends up successfully treating one
Bhagat and Welch (1995) find that higher or more diseases. After discovery, regulatory
debt and tax payments in the prior year issues pose additional risks. Even if a new
reduce current period R&D expenditures. drug successfully treats a disease, regulatory
In addition, Giaccotto, Santerre, and Vernon bodies such as the US FDA and the EMA
(2005) find that R&D spending increases decide whether the incremental clinical effi-
with real drug prices. They estimate the elas- cacy of a new drug, balanced against poten-
ticity to be around 0.6. tial safety considerations, merits approval.
The literature has, so far, not clearly dis-
3.4.1 The Determinants of Scientific
tinguished between a “capital constraints”
Productivity
explanation and an explanation emphasiz-
ing the use of current profits as a proxy for As mentioned earlier, fewer than 30 per-
future profits. For example, if firms with cent of drugs entering phase I end up sur-
higher sales today are inferring greater prof- viving until phase III (Adams and Brantner
its in the future, one may obtain a cash flow– 2006). What are the economic factors that
investment relationship even without capital influence success or failure?
market imperfections. However, there is evi- Cockburn and Henderson have written
dence that transient deviations from trend several papers investigating the relationship
profits predict R&D expenditures (Scherer between firm size and the productivity of
2001). Only capital constraints could explain innovation investments. In a 1996 paper, they
this finding, not e xpectation formation. This show that larger pharmaceutical firms enjoy
is not a perfect test, though, since one can- greater success, with both scale and scope
not rule out the possibility that current sales being possible contributors to that success
affect future expectations, as well. (Henderson and Cockburn 1996). Larger
Hall and Lerner (2011) present a com- firms are more successful, because they con-
prehensive review of the issues surrounding duct larger-scale research in any given areas,
financing and R&D. Somewhat intuitively, and because they conduct research in a wider
they conclude that capital market imperfec- variety of areas. Cockburn and Henderson
tions are most compelling for small and new (2001) produced a subsequent paper finding
firms. The evidence for larger firms is mixed, scope to be quantitatively more important
although they do note the revealed prefer- than scale. The latter suggests that external
ence of many large firms for using internal effects and knowledge spillovers within a
funds. In sum, the link between cash flow firm influence the productivity of research
and R&D is well substantiated in a number investments more than the sheer size of a
research enterprise. Simply increasing the an alliance partner. These results emphasize
scale of a particular project does not lead depth, rather than scope, in contrast to the
to greater productivity of innovation invest- Cockburn and Henderson (2001) study on
ments. However, pursuing a diverse port- earlier data.
folio of projects side-by-side does generate The studies above tackle the question of
greater productivity, hypothetically because how firms transform research inputs into
of knowledge being shared across adjacent discoveries. There is, however, another side
projects. of the market that has received much less
However, more recent work by Danzon, attention until recently. Malani and Philipson
Nicholson, and Pereira (2005) challenges (2011) analyze the supply of participants in
this view. Cockburn and Henderson (2001) clinical trials. They make the salient obser-
used a data set containing information from vation that this market forges a link between
ten pharmaceutical firms covering the years the utilization of current medical technolo-
1961–90. Danzon, Nicholson, and Pereira
gies and the development of future ones. In
note that this period largely predates the particular, if current medical technologies
arrival of biotechnology firms, and the ascent are of high quality, the relative returns to
of small- and medium-sized firms in biotech- entering a clinical trial of a novel technology
nology and pharmaceuticals. The success of with uncertain efficacy fall. Thus, the supply
such firms suggest the possibility of growing of trial participants will be greatest when
research success for small firms. It also pre- the quality of current-period technologies is
dates the arrival of contract research organi- poorest, and v ice versa. This implies, among
zations (CROs) that allow firms (particularly other things, that medical progress becomes
small ones) to outsource significant portions its own enemy, in the sense that rapid prog-
of their clinical R&D programs. In earlier ress discourages the clinical trial participa-
years, a small firm might have been hard tion that serves as the lifeblood of future
pressed to build the infrastructure necessary progress. As a result, medical progress is
to support a major, c ross-country clinical trial partially self-limiting, because technological
on its own. Today, it can outsource this activ- progress makes it harder to recruit clinical
ity to one of several highly experienced CROs trial participants and test future innovations.
that conduct these types of trials regularly.
3.4.2 The Effect of Regulation on the
All these trends mitigate the disadvantages
Quantity and Quality of Innovation
faced by small firms and suggest the possi-
bility that the returns to scale and scope may Regulatory factors play a significant role
have changed over time. To test this hypoth- in the probability of bringing a new drug
esis, Danzon, Nicholson, and Pereira study to market and, arguably, the probability
data on over 900 firms covering the period that it is safe and effective. In the United
from 1988 to 2000, largely after the period States, the FDA is responsible for approv-
studied by Cockburn and Henderson (2001). ing drugs based on the standard that new
In this newer data, Danzon, Nicholson, and drugs must be considered safe and effective.
Pereira find no evidence of scope econ- From an economic perspective, the cost of
omies, and even some evidence of scope the approval process is the additional time it
diseconomies related to the probability of takes for successful drugs to be launched and
bringing a new drug through phase III. They any social losses from beneficial drugs that
also find that a firm’s experience in a thera- are not approved. The benefit is the reduc-
peutic area increases the probability of phase tion in the use of drugs that are unsafe and
II and III successes, as does the presence of ineffective. The obvious question is whether
the benefits outweigh costs, and if not, what review becomes murkier. On the other hand,
reforms would lead to a more cost-beneficial since 56,000 life-years lost is a fairly extreme
process for launching new drugs. upper bound that assumes all deaths were
One aspect of this question is whether the consequence of PDUFA, perhaps even
increases in review times lead to the a small or zero net benefit should be viewed
approval of safer or more effective drugs. as making a case for faster review. They con-
More time and effort presumably lead to clude that FDA review times are too long
better decision making, but they also delay from a social perspective, because PDUFA
the launch of approved drugs. One study increased net social surplus on the margin.
that assesses the marginal costs and benefits A potentially more damning argument
of lengthier review is Philipson et al. (2008). against the benefits of longer FDA review
They use the implementation of the PDUFA times is made in a later paper by Grabowski
as a natural experiment. PDUFA was first and Wang (2008). They find that the length
passed in 1992. As its name implies, it gives of real-world experience is a much better
new drug applicants the opportunity to pay mechanism for ensuring safe use than is
a fee in exchange for faster review (though lengthy FDA approval. On the one hand,
not necessarily approval) of their prod- they demonstrate that drugs that are more
ucts. Philipson et al. quantify and value the novel, have more black-box warnings, and
increase in the speed of launch, and compare have shorter lag times between foreign and
this to the estimated cost of errors made due US launches, exhibit more adverse events.
to accelerated (and presumably less compre- These results are fairly intuitive, since
hensive) drug reviews. To quantify the mar- these drugs are presumably less under-
ginal value of PDUFA’s faster approval times, stood because they suffer from a lack of
they calculate the increase in firm profits real-world clinical experience. Yet, signifi-

due to faster launch using data on the time cantly, once these factors are controlled, the
series of revenues, the time cost of money, length of FDA review time does not predict
and the estimated increase in launch speed. adverse events. This suggests that longer
They then extrapolate the change in profits FDA reviews do not contribute to greater
to an estimate of the associated change in safety, once real-world experience is prop-
consumer surplus, using an estimate of the erly accounted for. Based on this finding,
fraction of social surplus appropriated by Grabowski and Wang conclude that more
innovators. Finally, they construct an upper resources need to be devoted to real-world
bound on the cost of errors due to faster surveillance, rather than lengthy agency
review times by assuming that all deaths reviews.
due to adverse events associated with drugs The earlier papers study the question of
launched after PDUFA were due to PDUFA. whether longer FDA review times lead to
Using this method, they find PDUFA gen- higher quality drugs. Some have asked more
erated a gross increase in consumer surplus radical questions that cut to the heart of the
due to faster launches of $14–$31 billion FDA review process. Manski (2009) has
and a maximum of 56,000 lost life-years due argued that the FDA process is fundamen-
to errors in approval. They point out that tally flawed in its insistence on either fully
if each life-year is worth $100,000, this is a approving or fully rejecting drugs. Instead,
good deal for society. Of course, if life-years he has suggested the possibility of adaptive
are worth $250,000, which is not implausi- partial approval. According to Manski, the
ble based on the literature (Hirth et al. 2000, central problem faced by the regulatory
Murphy and Topel 2006), the case for faster authority is not how much time to invest in
review, but the quality of data they are pre- c overage may be reversed in the future or
sented with. The “all-or-nothing” nature of allowed to stand (Mohr and Tunis 2010). The
approval means that innovators are encour- pure and ideal solution would be to allow
aged to complete approval trials quickly. As a access for some random sample of eligible
result, they are frequently unable to run tri- patients. Recognizing the practical problems
als long enough to study relatively rare, but associated with this, Manski proposes instead
important, outcomes such as deaths, heart that innovators be granted limited-term
attacks, strokes, and so on. Instead, they are licenses to sell a fixed quantity of their
encouraged to study so-called “surrogate drug based on prelaunch trial data. These
end points” such as low-density lipopro- licenses can be expanded as the evidence
tein (LDL) cholesterol, glycosylated hemo- base matures and improves, or rescinded
globin (HbA1C) for diabetes medications, if later studies demonstrate safety issues or
progression-free survival for cancer thera-
lack of effectiveness. From both economic
pies, and so on. Surrogate end points are cor- and statistical perspectives, this is an intui-
related with the “hard” outcomes of interest, tive and sensible idea. There is, however,
but imperfectly. In addition, clinical trials are the practical question of how markets would
conducted with relatively small samples of respond to the arrival of a quantity-limited
patients, and with n onrandom samples. Even license. In cases where the quantity restric-
though trials randomize patients in theory, tions imposed by the license failed to bind,
sample selection is made nearly inevitable, the “partial approval” would be equivalent
because patients must volunteer for trial par- to full approval and thus of little incremen-
ticipation and always retain the option of leav- tal value. In cases where it did bind, it is
ing the trial if they so choose. For all these not clear how markets and clinicians would
reasons, Manski argues that the FDA is oper- allocate the drug. For example, would clini-
ating with highly imperfect information, and is cians now need to develop priority scores for
thus subject to both type I and type II errors. which patients get the drug first? This would
Type I errors can theoretically be remedied, seem to create an additional layer of uncer-
because a drug later shown to be unsafe can tainty, as this prioritization decision itself will
be withdrawn from the market or have its use be based on limited information about which
limited. In contrast, type II errors are more patients benefit the most from the drug. One
insidious and potentially permanent. potential solution would be to focus on the
Manski (2009) observes that type II concept of a full but term-limited “tempo-
errors are made even more prevalent by rary approval” rather than an approval that
the discontinuous nature of approval deci- is both partial and temporary. This would
sions. He imagines instead that drugs could mimic current approaches to coverage with
be partially approved based on limited evidence development, and would avoid
information, and later have their approv- the prioritization problem. Such a strat-
als expanded or restricted, as a function of egy would only be useful, though, if it were
evidence that emerges later. While novel in accompanied by a commitment to approve
the drug approval context, this idea is not more drugs and thus mitigate type II errors.
without r eal-world precedent. Medicare, for The latter point is perhaps the most
example, already pursues an analogous strat- uncertain, because it cuts against the incen-
egy referred to as “coverage with evidence tives of drug approval bodies. Viscusi and
development.” It provides coverage for a Zeckhauser (2015) have emphasized the
new technology in the interim, while it col- cognitive and behavioral biases exhibited by
lects additional data on effectiveness. This FDA decision making that could complicate
the implementation of the otherwise sensi- second layer of economic review by indi-
ble approach proposed by Manski (2009). vidual countries that attempt to determine
They point out at least two types of biases. whether the benefit of the drug is worth its
First, the FDA appears to overweight type I cost. We take up this issue of reimbursement
errors relative to type II errors. Presumably, later, but it does suggest that the FDA plays
this is because the FDA itself is more likely a potentially more important role in patient
to be blamed for type I errors: it is difficult access to medications than many of its over-
to point the finger at a regulator who fails to seas counterparts.
approve a drug that would have been benefi-
3.5 Critiques of the Simple Theory and
cial. Second, the FDA is likely biased against
Their Normative Implications
drugs that have uncertain benefit, even if the
expected benefit justifies use. Such an aver- We now turn to the normative implica-
sion would also cut against a core principle tions of the simple theory for innovation and,
of adaptive approval, namely that the devel- in particular, to critiques of these normative
opment of evidence over time should be implications. The normative implications of
exploited as a means of gradually reducing the simple theory follow from a few basic
uncertainty. Current FDA behavior would principles. Define E(CS | D)as expected con-
lead to the outright rejection of a drug with sumer surplus in the discovery state, and
considerable uncertainty. E(CS | N)as expected consumer surplus in
In sum, there is significant evidence that the no-discovery state. The simple theory
FDA review times might be excessively long, then implies the following social planner’s
and that longer review times are not associ- problem:
ated with the approval of safer drugs. One
implication of this is that the current system    p(I)E( CS|D) 
max
I
might overweight the importance of FDA
review and underweight the importance of + (1 − p(I))E( CS|N)  − (1 + r)I.
gathering data on drugs after their launch.
This conjecture is fleshed out by Manski’s This has the first-order condition:
idea of adaptive partial approval, an emi-
nently sensible approach that must none- p′(I)[E(CS|D)  − E(CS|N)] = (1 + r).
theless cope with apparent behavioral biases
among drug approval and regulatory bodies. By comparing this first-order condition to
In terms of our overall framework, the liter- the private condition for optimal investment,
ature described here suggests that the FDA we see that efficiency obtains so long as:
might be reducing the productivity of inno-
vation investments by creating misalignment ϕ(I) = I
between drug approvals and the true social
value of drugs. E(π|D)  − E(π|N) = E(CS|D)  − E(CS | N).
The discussion here has focused on the
US context exclusively. There are meaning- The first condition implies that the innova-
ful differences across countries, however. tor bears the full social cost of investment in
For example, in Europe, drug approval is the first-best scenario. The second implies
typically a two-stage process. The EMA per- that she internalizes the full social benefit.
forms an initial review of drug safety and effi- Specifically, the latter condition requires that
cacy. Subsequently, drugs that are approved the incremental profit due to discovery is
by the EMA often still have to undergo a equal to the incremental consumer surplus.
This also requires that the innovator can fully While this might appear to be efficient, the
appropriate the entire consumer surplus asso- problem is that u nder-provision of the good
ciated with the innovation. This represents ex  post implies that E(CS|D)  − E(CS | N)
one of the strongest normative implications is also too low. Therefore, at best, we can
of the simple theory: investments in innova- achieve a level of innovation that is optimal
tion are efficient only if firms can appropriate conditional on an inefficiently low level of
the full value of their inventions, leaving con- consumer surplus. Policy approaches that
sumers with no surplus of their own. tackle both problems simultaneously are
In the absence of perfect p rice- thus preferable.
discrimination, innovators will harvest less These arguments rely on the prediction
than the full consumer surplus associated that innovators cannot capture the full value
with their invention. This leads to ineffi- of social surplus. There is support for this
ciently low levels of investment in inno- contention in the literature. Nordhaus (2004)
vation. A monopolist with access only to a examines a range of innovations—not just in
linear price will suffer from this inefficiency, the pharmaceutical industry—from 1948 to
which is made even more acute in the case of 2001. He concludes that innovators capture
preventive health investments, like vaccines, 2.2 percent of total present value to society
that are sold to healthy people (Kremer and of their inventions. A study more specific
Snyder 2006). As Kremer and Snyder have to health care is Philipson and Jena (2006).
argued, vaccines are sold before consumers They show that health-care innovators tend
know their illness status. Therefore, when to appropriate between 5 and 10 percent of
consumers vary in their risk of acquiring the total social surplus associated with their
the disease, there may be even more het- products.
erogeneity in demand for vaccines than for Few argue that health-care innovators are
treatments of the disease in question. This able to perfectly price discriminate or appro-
heterogeneity makes it even more difficult to priate consumer surplus fully. However,
appropriate total consumer surplus because several critiques of the simple theory of
the ideal of perfect p rice discrimination innovation have argued against its normative
becomes harder to achieve. implication that full appropriation of con-
Since innovative goods are often provided sumer surplus is an ingredient for efficient
by patent monopolies, the u nder-provision innovation investment. Much turns on the
of new ideas is often accompanied by the outcome of this debate. If one believes inno-
under-provision of innovative goods that
vation is under-provided, then reimburse-
embody new ideas. The simple theory links ment for new medical technologies ought to
these two problems, because one cannot be made more generous. On the other hand,
obtain fi
rst-best innovation in the presence critics of drug pricing often assert that there
of second-best goods provision. To pursue is too much low-value innovation being pro-
this point, consider a subsidy for innovation duced (Howard et al. 2015).
investment in the presence of second-best
3.5.1 Patent Races
goods provision. One might think that the
optimal policy would be a subsidy such that: One objection to the simple normative
theory revolves around the idea of “patent
ϕ′(I) < 1
races,” which criticizes the assumption of
a single firm undertaking research in iso-
E(π|D)  − E(π|N)
________________
  
      = E(CS|D)  − E(CS | N). lation. When multiple firms can enter into
ϕ′(I) a race to produce a given innovation, each
new entrant reduces the probability that any harmaceutical industry is somewhat uncer-
p
given incumbent firm will “win” the race to tain. Clearly, innovative firms directly com-
invent. This negative externality, which is not pete to reach the market first. However, even
internalized by the entrant, leads to exces- the “loser” ends up on the market, producing a
sive entry and excessive innovation invest- slightly differentiated product. For example,
ments. From an efficiency standpoint, this Gilead Sciences and AbbVie both launched
needs to be weighed against the incomplete novel therapies to treat hepatitis C infection.
extraction of consumer surplus, which leads Gilead was first to market. However, since
to insufficient entry and investment (Menell the two companies were working on slightly
and Scotchmer 2007). In the presence of pat- different, albeit related, therapies, both
ent races, therefore, it is not clear whether ended up in the marketplace. Gilead likely
innovation is too high or too low. obtained some advantage from winning the
Since entry creates a negative externality, race to enter, since monopoly prices are typ-
it follows that rewards for each entrant would ically higher than duopoly prices. However,
be socially excessive under perfect price dis- the advantage was not absolute. More gener-
crimination. For this reason, it is efficient for ally, pharmaceutical firms are rarely racing to
a successful innovator to receive less than launch fully identical products. More often,
the social surplus created by her invention. they race to launch similar but distinct mole-
However, it is unclear whether entrance and cules. There is likely a first-mover advantage,
innovation investments are socially exces- and thus a negative externality from entry.
sive or inadequate on balance, because the However, its magnitude may not be as sub-
negative entry externality competes with the stantial as in cases where losing a patent race
incomplete extraction of c onsumer surplus. shuts a firm out of a market entirely. Indeed,
The innovator’s ability to extract surplus in some cases, firms that are second or later
may be so poor that, even in the presence to market end up winning the greatest mar-
of the negative entry externality, there is ket share. Pfizer’s launch of Lipitor for the
insufficient incentive for firms to invest in treatment of cholesterol is one such exam-
innovation. ple: Pfizer was not first to market with a drug
In the patent-races model, the efficiency or in the statin class, but its statin was widely
inefficiency of market outcomes depend on viewed to be superior to its competitors in
the costs and benefits of duplicating research the class (Davidson et al. 1997).
efforts. For instance, if all firms succeed or The empirical literature paints a similarly
fail together, there is no incremental social complex picture of how patent-races models
benefit to the entry of a new firm; if suc- apply to the pharmaceutical industry. Studies
cesses and failures are totally uncorrelated, in other industries suggest that firms consis-
however, the opposite is true (Loury 1979, tently responded to changes in rivals’ R&D
Lee and Wilde 1980). A number of more policies (Grabowski and Baxter 1973, Lerner
subtle issues may also arise in the analysis of 1997). However, Cockburn and Henderson
patent races. Inefficiencies can be created (1994) search for and fail to find similar
by asymmetric information, whether this is patterns in data from the pharmaceutical
about c ost efficiency, the value of innovation, industry. They hypothesize that firm-specific
or the state of technical progress (Scotchmer differences in research productivity and
and Green 1990; Bhattacharya, Glazer, and costs of quickly adjusting research spending
Sappington 1992). may play a larger role in the pharmaceuti-
While the patent-races model is intui- cal industry and complicate the application
tive and appealing, its applicability to the of standard patent-races models. This cuts
against the prediction of excessive entry into of the incumbent and make it harder for
pharmaceutical research and development. follow-on inventors to enter the market.

However, this discourages innovation invest-
3.5.2 Cumulative Innovation
ment by potential follow-on inventors, and
The example of Lipitor also highlights a thus may lead to inefficiently low levels of
substantively different objection to the simple investment by f ollow-on inventors. IP licens-
normative theory. The literature on “cumula- ing can help solve this problem, because it
tive innovation” critiques the assumption of allows incumbents to be compensated by
a single firm conducting research that “starts follow-on firms that add value to the orig-
from scratch” without building on earlier inal invention. Ideally, one can imagine
ideas. For example, if other firms research- schemes where profits are shared perfectly
ing drugs in the statin class benefited from between incumbents and followers such
the work done by Merck on Mevacor (the that the follow-on firms receive an amount
first statin launched), free riding or other exactly equal to the incremental social value
obstacles to efficiency may emerge. they create. When firms share profits, opti-
The simple model conceives of a firm mal patent length rises in order to offset
producing a single innovation, which exists the decline in profits per period (Green and
independently of other past or present dis- Scotchmer 1995).
coveries. In reality, new discoveries depend In practice, such licensing arrangements
on previous inventions that unlock the door are relatively uncommon in the pharma-
to future inventions. This cumulative process ceutical industry. As discussed earlier,
poses a number of incentive problems. On companies that are first to market with a
one hand, inventors may lack the incentive to drug exhibiting a new mechanism of action
stimulate follow-on research by competing receive a first-mover advantage. However,
firms. In this sense, cumulative innovation they receive no other compensation from
exacerbates the problem of underinvestment follow-on firms that subsequently enter the
in innovation. On the other hand, follow-on market with distinct drugs using the same
inventors may be motivated by the pros- mechanism of action. Indeed, follow-on
pect of “capturing” the market held by an innovators may even receive the lion’s share
incumbent. This creates excessive incentive of the value, if their drugs are superior in
to innovate, since the new entrants are cap- quality to that of the original entrant. The
turing value that the incumbent has already case of statins discussed earlier presents one
created. In the context of pharmaceuticals, example. The first statin to lower cholesterol,
for example, consider a new drug that con- Mevacor (lovastatin), ended up as inferior to
sumers will pay $10,000 for, compared to later statins, in terms of the ability to reduce
an incumbent that they will pay $9,900 for. cardiovascular risk and mortality (Davidson
The new drug is only slightly better than the et al. 1997). The statin market was ulti-
old, but its introduction might allow its pro- mately dominated by the clinically superior
ducer to capture the entire market and thus follow-on drugs.
earn revenues of $10,000 per patient. This Perhaps as a result of this issue, some have
is true even though it only created $100 of argued that there is insufficient incentive for
incremental value per patient. The new firm pharmaceutical firms to invest in truly novel
may enjoy a partially free ride, courtesy of drugs (Bach 2014). The first entrant will not,
the value created by the original innovator. in general, be compensated for unlocking
One solution to the “excessive capture” a new class of drugs and thus there may be
problem is to strengthen the patent rights insufficient incentive for opening up a new
class. Once again, however, it is unclear Even without perfect price-discrimination,

whether this issue offsets the problem that subsidies provide innovators with the oppor-
innovators might harvest very small propor- tunity to harvest the cash value of the subsidy,
tions of social value, well below what even in addition to the value of consumer willing-
the negative externalities might suggest. ness to pay (Lakdawalla and Sood 2009).
On the other hand, insurance provides an
3.5.3 Implications of Health Insurance
opportunity to mitigate or eliminate dead-
The simple theory of innovation applies weight losses due to patent monopolies and
most directly to standard goods markets provides the innovator with opportunities to
in which consumers pay directly for their appropriate a greater share of consumer sur-
purchases. Health care differs, due to the plus. Health insurance breaks the usual link
presence of health insurers. Consumers pay between patent protection and deadweight
premiums to health insurers, in exchange for loss from monopoly pricing. With insurance,
reduced prices paid to health-care providers. it is possible to pay monopoly prices to the
In turn, insurers negotiate reimbursements innovator, but still charge low (or at least
with health-care providers, including phar- lower) prices to the consumer. Insured con-
maceutical firms. These institutional details sumers pay only a copayment or coinsurance,
complicate the application of the simple which will be lower than the price paid by
theory of innovation to the pharmaceutical the insurer to the manufacturer. This logic
market. has implications for both public and private
On the one hand, a variety of regula- insurance programs.
tions and insurance market imperfections Within a public insurance program, policy
inflate the market w illingness to pay for makers have the freedom to configure copay-
pharmaceuticals. Under perfect efficiency ments and reimbursement rates in a socially
with frictionless and unsubsidized health optimal fashion. The government can set fixed
insurance markets, consumers would be copayment levels in order to drive consumer
paying premiums that reflect their ex ante prices toward socially efficient marginal cost
willingness to pay for health care products pricing, even when manufacturers receive
(Lakdawalla and Sood 2013). However, monopoly reimbursement levels from the
consumer health insurance purchases are insurance scheme. The result is an improve-
subsidized through various mechanisms, ment in static efficiency without compromis-
including direct public provision of insur- ing dynamic incentives to innovate. Premium
ance—e.g., Medicare and Medicaid—or tax payments, government spending, or some
exemptions for employer-provided health combination of the two cover the difference
insurance premiums. These subsidies cre- between consumer prices and manufacturer
ate a wedge between private willingness to reimbursements. Provided that the innovator
pay for products and the w illingness to pay does not get overcompensated for his efforts
for an insurance contract that covers the as a result, insurance improves both static
costs of those products. In theory, a perfectly and dynamic welfare by limiting deadweight
price-discriminating innovator could capture loss and encouraging innovation (Lakdawalla
consumer surplus from his invention plus and Sood 2009).
the value of all subsidies for health insur- Private insurance markets can also limit
ance. This incentive leads toward excessive deadweight losses from patent monopolies.
innovation and counteracts the problem of Health insurance contracts typically take
imperfect consumer surplus appropriation the form of an ex ante premium p ayment,
by firms (Garber, Jones, and Romer 2006). coupled with an ex post unit price for the
consumer. This resembles the standard might improve welfare. One caveat here
“two-part pricing” contract that allows firms concerns the potential impact of insurance
with market power to extract profits without benefit design features. If insurers design
constraining quantity and triggering the cor- benefits around a percentage coinsurance,
responding deadweight losses (Lakdawalla rather than a flat copayment, innovators
and Sood 2013). Notably, this is only true for could potentially mitigate or even offset
insured consumers. Uninsured consumers the effects of insurance by raising their own
are exposed to the full monopoly loss asso- price. For instance, if insured consumers
ciated with patents, and they pay monopoly pay 20 percent of the market price, insur-
prices for pharmaceuticals and other medical ance may, in theory, allow firms to inflate
innovations. Therefore, lowering the rate of their prices by as much as a factor of five.
uninsurance also has the desirable side effect As a result, consumers might not enjoy price
of limiting deadweight losses due to patent reductions, and the manufacturer would
power in health care (Lakdawalla and Sood capture all the rents from the insurance sub-
2013). sidy without any improvement in efficiency.
This argument embeds another signifi- The government can avoid this by regulating
cant point: competition among health-care the cost-sharing schedule offered to manu-
providers benefits consumers, and market facturers. Drug insurance will improve static
power of all kinds harms them. While these efficiency if, at the monopolist’s margin,
points may seem obvious outside health care, increases in the manufacturer price impose
for decades, many health economists argued greater out-of-pocket costs on consumers.
that health-care competition reduces welfare Under this condition, a price increase will be
by exacerbating moral hazard that results less well-insured than the initial price itself.
from low out-of-pocket costs for consum- In practice, it is fairly common for private
ers (Crew 1969; French 1996; and Folland, insurers to pass through some fraction of
Goodman, and Stano 2001). Monopoly was drug price growth to consumers; this con-
thought to be a second-best improvement dition is thus likely to be met in real-world
in welfare because it boosted out-of-pocket markets (Lakdawalla and Sood 2009).
costs. Gaynor, H aas-Wilson, and Vogt (2000)
3.6 Implications for Public Policy Toward
demonstrated why this reasoning is flawed:
Research and Development
logically, if private insurers could make their
customers better off by raising copayments, The preponderance of evidence suggests
they would surely do so. The early litera- that raising reimbursements for pharma-
ture abstracted from this response of insur- ceuticals stimulates innovation, primarily
ance contract design to medical care prices. because the expected rewards for innova-
Incorporating this response reverses the tion go up and secondarily because the cost
implications of market power for welfare. of financing falls for c ash-constrained phar-
In sum, competition is good for consum- maceutical firms. This positive implication
ers, and market power is bad. Moreover, pri- remains challenging to translate into clear
vate health insurance limits the social cost policy prescriptions, however, because it
of market power due to patent monopolies remains unclear whether pharmaceutical
and other institutions. In principle, this helps innovation is too high or too low. The canon-
build a case for subsidizing privately provided ical model would imply chronic underinvest-
prescription drug insurance. If reducing the ment in innovation, but a variety of more
rate of uninsurance lowers deadweight loss, complex alternatives—e.g., patent races—
subsidies aimed at achieving this outcome lead to ambiguous implications.
Yet, there are plausible paths forward that be much larger—for example, if each year of
may be used to recover normative conclu- life were conservatively valued at $100,000,
sions. To a first approximation, the benefits reducing life expectancy by six months in
of innovation investment can be measured the United States would result in losses that
as years of life gained or q uality-adjusted exceed GDP. This type of reasoning has been
years of life gained. The costs, on the other used to demonstrate that the potential ben-
hand, can be approximated as the additional efits of lowering prescription drug prices are
spending on new medical innovations. If in relatively modest in size compared to the
fact innovation is socially inadequate, then potential losses that may result from insuf-
an extra dollar of R&D investment should ficient innovation (Lakdawalla et al. 2009).
generate less than a dollar of net social value, Thus, in spite of the gaps in the evidence,
which is defined as life-years gained net of rational policy makers may see a bet on stim-
additional spending on innovation. There has ulating innovation as a less risky proposition
been little direct empirical research shedding than its reverse. While useful and practical,
light on this question, although modeling this analysis should not be seen as a substi-
approaches have been taken that combine a tute for rigorous empirical quantification
range of parameter estimates from the liter- of the social value of greater innovation
ature to arrive at a conclusion. At least one investment.
of these suggests that stimulating innovation More evidence may be needed on the
generates positive social returns (Lakdawalla value of stimulating innovation in developed
et al. 2009), but any modeling exercise is markets, but a small but highly effective lit-
hampered by uncertainty about the under- erature has made great headway on guiding
lying parameters. A single, w ell-designed innovation policy as it relates to emerging
empirical strategy would be much more markets. Kyle and McGahan (2012) find
definitive, but such an approach has so far that patent protection in emerging markets
proven to be elusive. has little to no effect on innovation into dis-
In spite of the uncertainty, innovation pol- eases that afflict these markets. At the same
icy must get made. Rational policy makers time, Chaudhuri, Goldberg, and Jia (2006)
must make decisions that incorporate uncer- find that patent protection in these mar-
tainty about the social consequences of stim- kets imposes considerable welfare losses
ulating innovation. A r isk–reward calculus is on consumers, ranging from 25 percent to
needed. It is worth noting that the costs of 75 percent of total spending. Chaudhuri
over-innovation are not necessarily symmet- Chaudhuri, Goldberg, and Jia focus on one
ric with the costs of u nder-innovation. Too particular class of antibiotics (quinolones)—
much innovation results in excess spending in a single, albeit, important emerging mar-
on drugs. The welfare loss can be quantified ket (India). Thus, perhaps confirmatory
as the excess spending on drugs and on inno- evidence is needed for other countries and
vation itself. Since excess spending on drugs drug classes. Nonetheless, taken together,
is bounded above by total spending on drugs these two papers alone reach the striking
(which currently sits at about one-sixth of conclusion that patent protection in emerg-
total health-care spending), the total down- ing markets produces little if any dynamic
side risk associated with overstimulating efficiency improvement but significant static
innovation faces a practical and quantifiable welfare loss. This raises the l ong-run question
limit. In contrast, too little innovation results of how to manage intellectual property pro-
in lives lost, which is much harder to bound. tection during an economy’s transition into
The potential economic consequences could a developed state. The risk is that emerging
markets become locked into an equilibrium data set of detailed R&D spending data
with limited patent protection, and then face that allows them to construct and release
the difficult task of imposing protection and estimates of research costs per launched
its associated costs on consumers when their drug. However, there is no way for other
economies are much larger. researchers to replicate these data, or to

The latter point also illustrates the complex obtain disaggregated details. This has ham-
political economy of IP protection. Emerging pered research on the cost of the marginal
markets, just like each individual country, drug to society, although we do have esti-
face a free-riding problem because innova- mates on the cost of the average drug.
tion is a global public good. Each country On the other hand, the value of the mar-
would prefer that other countries protect IP ginal drug is hampered by the difficulty of
and stimulate global innovation. In practice, measuring health outcomes. The one uni-
the United States currently makes up about form way of measuring benefits for drugs
half the global pharmaceutical market. Its is to compute the gain in quality-adjusted
own policies thus have an outsized impact life-years (QALYs) associated with treatment
on global innovation. Restrictions in pat- by the relevant drug. QALYs incorporate
ent protection or in the rewards that can be both gains in life expectancy and in qual-
earned within the United States would likely ity of life. However, QALYs are not always
have meaningful negative effects on rates of measured in individual-level databases.
innovation, and thus health outcomes, for Thus, conventional approaches of identify-
patients all over the world (Lakdawalla et al. ing exogenous variation in drug launches and
2009). Conversely, many US policy makers connecting these to changes in QALYs at
have expressed the concern that the United the population level are not always possible.
States is now footing the bill for global inno- No other approaches have emerged to fill
vation. Indeed, Egan and Philipson (2013) this gap. However, there is some hope for
have pointed out that larger countries may be solving this problem. A few longitudinal
expected to pay higher prices. Moreover, they patient-level databases—notably the Health
point out that the rise of emerging markets and Retirement Study (and the Medical
might reduce global returns to innovation Expenditure Panel Survey)—contain infor-
by exacerbating the free-riding incentives mation that allows researchers to compute
faced by today’s large consumers, such as the patient-level QALYs. What remains missing
United States. Short of coordination or con- is an elegant identification strategy to bring
solidation across sovereign nations, there is to these data. Another challenge is the rela-
no obvious way of mitigating these f ree-rider tively small sample size of longitudinal data-
problems on the horizon. bases—typically under 10,000 patients—and
the relatively small p
opulation-wide effect of
3.7 Gaps, Limitations, and Unanswered
a single drug launch. The solution may entail
Questions
incorporating QALY measures into very
Perhaps the biggest unanswered question large databases, such as Medicare claims
about pharmaceutical research is whether or other similarly large-scale repositories of
we are doing too much, enough, or too lit- health-care information.
tle. Thus far, the value and the cost of the
marginal drug to society remain unknown.
4. Pricing Strategies
Detailed data on R&D costs are scarce.
The Tufts University Center for Drug The pervasiveness of p
atent-driven market
Development has obtained a confidential power within the pharmaceutical industry
gives rise to a wide variety of pricing arrange- To shed light on this framework, consider
ments, as pharmaceutical firms experiment an example formulary dictating that patients
with various approaches to harvest the great- pay $5 per prescription for generic drugs,
est share of value from their customers. At $30 per prescription for “preferred” branded
the same time, monopsony power exerted drugs, and 20 percent of the drug cost per
by powerful public payers and private payers prescription for “ non-preferred” branded
creates countervailing downward pressure drugs. The differential out-of-pocket pric-
on prices. ing in a formulary is sometimes referred to
as “tiering.” Manufacturers that charge very
4.1 US Private Market Pricing
high prices to the insurer may be relegated
Unlike in most other developed countries, to a non-preferred tier charging patients
pharmaceutical pricing in the United States 20 percent of the drug’s cost. Patients may
is largely decentralized, albeit influenced defect from these drugs instead of paying the
by several powerful p ublic-sector buyers. It high cost. On the other hand, manufacturers
is also quite disintegrated, involving private that charge low prices to the insurer might
insurance companies, pharmaceutical manu- enjoy placement on a preferred tier with the
facturers, pharmacies, drug wholesalers, and flat $30 copayment. The formulary structure
pharmacy benefit managers (PBMs). provides manufacturers with an incentive to
The uninsured segment of the private charge lower prices to the insurer.
market is simplest to understand. Here, A few additional examples help illustrate
uninsured customers buy directly from retail the incentive effects of different formulary
pharmacies. Upstream, pharmaceutical designs. Suppose that a manufacturer of a
manufacturers sell their products at a largely highly novel, patent-protected drug enjoys
uniform price. This transaction may or may monopoly power and faces the demand
not be mediated by a wholesaler, depending curve, Q = A − bP . A simple formu-
on the size of the pharmaceutical firm and lary design would impose x% cost-sharing
the pharmacy. The wholesaler keeps very lit- on patients for this drug. In this case, the
tle margin and is a relatively passive player in out-of-pocket price becomes P × x%, and the
the marketplace (Lakdawalla and Yin 2015). demand curve rotates clockwise to reflect
The “passivity” of the wholesaler is consis- this percentage price reduction. This struc-
tent with the finding that standard monop- ture provides greater incentives for price
oly pricing models fit the uninsured market restraint than a flat copayment of $z, which
reasonably well (Lakdawalla and Sood 2013). corresponds to quantity of Q = A − bz
The standard monopoly or oligopoly pric- that does not vary with the manufacturer
ing model has also been extended to the price.
insured marketplace by a number of authors. In the case of oligopoly, we might have a
A useful summary of this approach is offered scenario where drug manufacturers submit
by Berndt, McGuire, and Newhouse (2011). bid prices to the insurer, who instructs each
First, an insurer decides on how to design a that only the lowest-bidding firm will obtain
copayment or coinsurance schedule for drugs, favorable formulary position. One useful pre-
referred to as its “formulary.” After this formu- diction of this theory is that formulary tiering
lary is constructed, a monopolist or oligopolist can trigger greater oligopolistic price com-
sets prices in the usual way. Finally, based on petition among manufacturers of distinct but
the pricing decisions, the insurer will decide patent-protected drugs. If the lowest bidder
on which drugs to cover and, if applicable, on gets preferential out-of-pocket pricing, the
the formulary position of each drug. incentive effects are obvious.
While these models provide useful pre- (iii) the reimbursement paid by the insurer
dictions, one limitation, as noted by Gaynor, to the pharmacy; and (iv) a dispensing fee
Haas-Wilson, and Vogt (2000), is that they paid by the insurer to the pharmacy.
overlook the endogenous nature of insurer Further complicating matters has been
formulary and contract design decisions. The the advent of PBMs. Traditionally, PBMs
“sequential” modeling approach presumes focused on processing and administer-
that formulary design influences manufac- ing prescription drug insurance claims.
turer pricing, but it overlooks the possibility However, they have also become involved in
that pricing influences formulary design too. the negotiation of prices and rebates in the
Gaynor et al. explain how this assumption marketplace (Scherer 2000, pp. 1325–28).
leads to erroneous inferences. For example, it Indeed, it seems natural for PBMs to nego-
implies that market power and higher prices tiate, since they are also setting formulary
for drugs can benefit consumers afflicted by design, which provides them with lever-
moral hazard, simply by lowering utilization. age against manufacturers. Unlike insur-
This c ounterintuitive result breaks down as ers, PBMs do not typically bear financial
soon as we allow for insurers to design ben- risk. They relieve the risk-bearing entity—
efits optimally. A rational payer would make whether a traditional insurance company or
consumers better off by raising out of pocket a self-insured employer—of the administra-
prices on its own, without needing he stimu- tive burdens associated with pharmaceutical
lus of outside market power (Lakdawalla and insurance. They allow specialization of firms
Sood 2013). across financial risk bearing and insurance
Price bargaining models can account for administration.
these and other anomalies in the sequential To avoid overcomplicating the economic
model. To understand how and why, it helps analysis, consider an insurer that internalizes
to review the institutional structure of the the two functions of risk bearing and admin-
pharmaceutical market and the way drugs istration. This insurer interacts with drug
are delivered to and paid for by insured con- manufacturers and pharmacies. Insurers
sumers. Just as in the uninsured context, simultaneously set the premium charged
manufacturers sell pharmaceuticals—pos- and the design of benefits offers to insureds.
sibly through a wholesaler—to a pharmacy, Thus, there are at least two components to
which then retails to consumers. Insured the economics of this marketplace:
consumers pay the pharmacy a copayment
when they purchase drugs, and their insur- (a) 
Two-part pricing of drugs, where
ance company reimburses the pharmacy an insurers set copayments and benefit
additional amount, intended to make up the schedules, and insurers bargain with
difference between the copayment and the manufacturers and pharmacies over
cost of the drug (in addition to a “dispensing prices and rebates
fee” that the pharmacy might also charge).
Finally, the insurance company may receive (b) 
Three-way bargaining over price,
a “rebate” from a manufacturer if it uses a among pharmacies, insurers, and man-
sufficiently large volume of the manufactur- ufacturers—possibly even more than
er’s product. Thus, at least four quantities three-way, if one involves wholesalers,
are simultaneously negotiated in the private PBMs, and other intermediaries
marketplace: (i) the price paid by the phar-
macy to the manufacturer; (ii) the rebate As a result of the first component, manufac-
paid by the manufacturer to the insurer; turers do not need to price on the consumer
demand curve because they are not sell- This finding is consistent with the notion
ing directly to consumers (Lakdawalla and that larger buyers enjoy more favorable pric-
Sood 2013). Instead, they need to price on ing terms in the pharmaceutical market. In
the competitive insurer’s break-even curve. addition, price reductions due to buyer size
Alternatively, they will bargain with insurers are more prevalent for generic drugs than
who have market power. branded drugs, consistent with the theoreti-
Theoretically, pricing within the privately cal result that pharmaceutical manufacturers
insured US market consists of bargaining will share more of the surplus with payers
among pharmaceutical firm and insurer (or and pharmacies when they are selling into
PBM) over rebates; pharmaceutical firm and more competitive markets (Lakdawalla and
pharmacy over drug acquisition costs; and Yin 2015). Intuitively, if manufacturers pos-
pharmacy and insurer over reimbursements sess all the market power—perhaps when
and dispensing fees. The typical approach to they sell a drug with no substitutes—insurers
bargaining models in this literature begins cannot extract surplus no matter their size.
with N ash bargaining of some form (Ellison At the other end of the spectrum, generic
and Snyder 2010), and a t hree-way Nash bar- drug manufacturers typically send substan-
gaining problem in particular (Lakdawalla tial portions of social surplus downstream to
and Yin 2015). pharmacies, insurers, and consumers.
Bargaining leverage across the three par- This complex structure of pricing within
ties depends on the availability of substi- the privately insured US population is not
tutes, the size of each player, and the relative very efficient. Most notably, pricing con-
importance of each player to the other. For tinues to be linked to the volume of drugs
example, a firm marketing a generic drug, or sold. Loosely speaking, drugs are priced by
a branded drug with many close substitutes, the gram or milliliter. This cost per unit of
will have a harder time striking generous quantity might be offset with rebates if cer-
pricing terms than one selling a novel drug tain volume targets are reached, but prices
with no close substitutes. The theoretical are still inherently linked to the physical
effects of insurer size are less clear. In gen- quantity of drug sold. This might seem sen-
eral, buyer size has ambiguous effects on the sible and straightforward when one is con-
equilibrium price in a bargaining problem, sidering a pill to treat high blood pressure,
because larger buyers may possess more but in many cases, the physical volume of a
leverage, but also suffer from inefficiencies drug is a poor measure of value to consum-
that reduce their attractiveness as a trading ers. A widely cited example is bevacizumab,
partner (Chipty and Snyder 1999). which was originally launched and priced as
Empirically, increases in buyer size within a treatment for metastatic colorectal can-
the US pharmaceutical market appear cer. It was later found to help patients with
to lower the price buyers have to pay. an eye disease called age-related macular
Lakdawalla and Yin (2015) used the imple- degeneration. Injecting 1.25 –2.5mg of bev-
mentation of Medicare Part D as a natural acizumab into the eye was found to be quite
experiment that expanded the size of private beneficial (Arevalo et al. 2008). In contrast,
insurers by subsidizing private prescription bevacizumab is used in much larger quan-
drug coverage. Insurers located in states tities—5 –10 mg per kg of body weight—in
with more Part D -eligible individuals grew the treatment of cancer. Since the drug was
for plausibly exogenous reasons, and were priced per milligram, and with the expecta-
able to extract better pricing terms for both tion that hundreds of milligrams would be
their Part D and n on-Part D beneficiaries. used at a time, ophthalmologists were able
to use bevacizumab to treat a ge-related mac- receive this much lower price. Thus, even
ular degeneration for pennies on the dollar. though public payers are considered less
However, it is hard to argue that bevaci- dominant in the US market overall, their
zumab generated orders of magnitude more behavior casts a long shadow over the entire
value for colorectal cancer patients than for private market.
macular degeneration patients.
4.2 US Public Market Pricing
The problems of q uantity-based pricing in
the pharmaceutical market have been much The vast majority of publicly financed
discussed, and a variety of alternatives have prescription drug purchases occur through
been proposed (Carlson et al. 2014). Some Medicare’s Part D prescription drug insur-
examples include: “ performance-linked ance program, and the Medicaid system.
reimbursement,” where a manufacturer Together, these two programs account for
is compensated as a function of clinical about o ne-third of the more than $300 billion
improvement in patients; or “conditional US pharmaceutical market (Schumock et al.
treatment continuation,” where payers con- 2014, Bruen and Young 2014, and Kaiser
tinue to provide (and pay for) a drug only Family Foundation 2014). Each program
if patients achieve p respecified clinical tar- interacts in unique and important ways with
gets in an initial period. To date, there have the private market for prescription drugs.
been a few halting steps toward uptake The 2003 Medicare Modernization Act
of these novel pricing schemes, mostly (MMA) established Medicare Part D, which
by public payers overseas. For example, began providing prescription drug cover-
Johnson & Johnson contracted with the age to Medicare beneficiaries in 2006. Part
UK National Health Service (NHS) on a D sets a standard prescription drug benefit
performance-linked reimbursement con-
level, then provides a subsidy worth at least
tract for the cancer drug, Velcade. Under 75 percent of the actuarial cost of this stan-
this agreement, Johnson & Johnson agreed dard benefit. L ow-income beneficiaries are
to refund the NHS for the first four months eligible to receive even higher subsidies.
of treatment cost for patients whose tumors While the original fee-for-service Medicare
failed to shrink by at least 25 percent after system features public administration of
initiating the therapy. insurance claims, Part D subsidizes the pur-
Interestingly, the private, decentralized chase of private insurance for prescription
US insurance market is innovating less rap- drugs. To be eligible for inclusion in the Part
idly in pharmaceutical pricing than its cen- D system, private insurers must provide a
tralized public peers overseas. This may have benefit that is actuarially equal to or better
more to do with the regulation of the private than the standard drug benefit. Once eligi-
market in the United States than with the ble, insurers enjoy the benefits of a subsi-
failings of private enterprise. As we discuss dized marketplace.
in the next section, pharmaceutical man- A number of authors have demonstrated
ufacturers in the United States are obliged that the passage of Part D stimulated the uti-
to sell drugs at the lowest-observed price to lization of prescription drugs, as one would
state Medicaid programs. Thus, if any novel expect (Duggan and Scott Morton 2010, Yin
pricing scheme drives down the unit price of et al. 2008, and Lichtenberg and Sun 2007).
a drug—say, because of poor performance in Duggan and Scott Morton (2010) also found
a particular subpopulation of patients—the that Part D appeared to decrease prices for
costs to the manufacturer are multiplied, drugs. They interpret this as the movement of
because Medicaid now might demand to individuals from uninsured status to insured
status, where they enjoy the bargaining As mentioned earlier, Medicaid rebate
power of a large buyer. Subsequent research rules—sometimes known as “ best-price”
finds that much of this pricing effect is actu- rules—also stifle innovation in private market
ally due to prices falling among the insured: pricing. Just as in any situation with market
Part D expanded the size of health insurers power, pricing variability can serve as a pow-
and thus led to greater bargaining power for erful tool for limiting deadweight loss in the
insurance companies. The result is lower pharmaceutical market. In competitive mar-
prices, even for consumers who remained kets, a uniform marginal cost-driven price is
insured before and after the passage of the efficient because all consumers who place
law (Lakdawalla and Yin 2015). more value on a product than it costs on the
From a normative perspective, prescrip- margin to produce will buy it. Under market
tion drug insurance subsidies accomplish the power, however, uniform pricing inevitably
twin goals of limiting deadweight loss in the excludes some consumers from the market
patent-protected pharmaceuticals market,
who place value on the good that exceeds
while simultaneously encouraging innovation cost. Price discrimination is uniquely possi-
investments (Lakdawalla and Sood 2009). In ble in the pharmaceutical market because
the absence of insurance, greater incentives opportunities for resale are severely limited.
for innovation mean higher current prices, Yet, Medicaid best-price rules limit incen-
thus greater deadweight loss. However, pre- tives to seek out price-discrimination strat-
scription drug insurance allows consumers egies. For example, consider a firm selling
to face lower out-of-pocket prices, even as a new cholesterol-lowering drug. This drug
drug companies earn higher revenues. To might lower LDL cholesterol by a dramatic
understand the latter point, note that utili- eighty points in some patients, but a mod-
zation rises in the wake of greater insurance, estly effective twenty points in other patients.
while prices typically do not fall by as much. The pharmaceutical firm can choose to price
Perhaps not surprisingly, the passage of Part its product to attract the moderate-value
D has been linked to greater profitability twenty-point patients or to attract the

in the pharmaceutical industry (Frank and high-value eighty-point patients. If there

Newhouse 2008, Friedman 2009). are enough high-value patients present, the
While Medicare Part D is widely viewed firm will choose to price the m oderate-value
as a boon to the pharmaceutical industry in patients out of the marketplace. A natural
the US market, Medicaid drug purchasing solution to this problem would be to charge
presents its diametric opposite. Medicaid prices that vary with LDL reduction. The
sets prices as a fraction of private market challenge is that some patients may happen
prices and then requires that manufac- to experience little or no LDL reduction and
turers issue rebates if those prices end up thus may pay very low prices. The record of
being higher than the lowest observed pri- even a small number of very low transactions
vate market price. This arrangement pro- prices awarded to a small payer in the private
vides pharmaceutical manufacturers with market may trigger automatic rebates to all
incentives to raise prices in the private state Medicaid programs. This has the effect
market so as to avoid paying large rebates of decreasing the expected rewards from
to Medicaid. Duggan and Scott Morton novel pricing strategies or other variable
(2006)estimate that a 10 percentage-point pricing techniques that could enable better
increase in Medicaid market share leads price discrimination.
to 7–10 percent increases in the price of a Public insurance programs also have
prescription. direct effects on revenues and utilization.
When individuals switch from private insur- this discrepancy may have been even greater
ers to Medicaid, they drive down average than Ellison and Wolfram measure, since
unit prices in the pharmaceutical market- they did not observe confidential rebates,
place, because Medicaid drug prices are only discounts negotiated on the unit cost.
weakly lower than private prices by statute The evidence suggests that firms may
(Duggan and Scott Morton 2006). The same respond to political pressure or the threat of
is true when uninsured people gain cover- regulation by lowering their observed prices.
age through Medicaid or Medicare Part D, The tangible benefits to consumers remain
because public or private insurers pay lower unclear, however, since the effects on actual
prices than the uninsured (Duggan and Scott market prices may be smaller.
Morton 2010). On the other hand, providing
4.3 Public Customer Pricing in Overseas
coverage to the uninsured may still boost
Markets
demand and thus raise overall utilization
in the marketplace. The effect of switching While the US pharmaceutical market is
patients from private payers to Medicaid far from unregulated, government influence
has ambiguous effects on utilization because over pricing remains indirect. In contrast,
Medicaid offers lower out-of-pocket pay- overseas markets frequently take a much
ments, but also makes it harder to access more direct approach toward government
certain novel drugs. involvement in pharmaceutical pricing.
Even the threat of regulation might influ- If drugs were freely resold in a global
ence private pricing behavior. If pricing marketplace, individual governments would
regulation is costly for legislators and regu- exert much less influence over domestic
lators to impose, the returns on those costs pricing. However, restrictions on reimpor-
will be higher when voters perceive greater tation of drugs serve as a brake on interna-
harms from high prices. Thus, higher prices tional arbitrage (Berndt 2002, pp. 60–61].
might heighten the risk of more regulation, These restrictions benefit low-income coun-
and vice versa. Ellison and Wolfram (2006) tries, which have a strong incentive to main-
assess this hypothesis using the health-care tain them. For example, suppose Greece
reform debates of the early 1990s as a type began to reexport large quantities of prod-
of natural experiment. They argue that firms ucts to h
igher-income Germany. Drug com-
with a greater pool of patent-protected sales panies would be encouraged to limit their
remaining, and firms more reliant on utili- sales to Greece in order to avoid damaging
zation by Medicare patients, are more vul- the more lucrative German market. Indeed,
nerable to pricing regulation. Thus, they the greater importance of Germany and the
argue that such firms would be more likely lesser importance of Greece place Greece
to invest resources in forestalling pricing in a vulnerable position that it might seek to
regulation. In support of this hypothesis, protect.
they show that the arrival of information The resulting segmentation of the global
about heightened risk of regulation caused market into individual country markets
the more vulnerable firms to curtail their provides governments with leverage over
list prices more than their peers. Notably, domestic pharmaceutical prices. In this
list prices were restrained more than actual environment, governments take a range
transaction prices in the market. Firms may of approaches to regulating pharmaceuti-
give discounts or rebates to some custom- cal pricing. Most fall into one of two broad
ers, and this may cause transaction prices to categories: reference pricing and price
depart from list or “sticker” prices. Indeed, controls.
Reference pricing stipulates that drugs will setting the reference price within a group.
be reimbursed at a level equal to the price of For example, the first reference-pricing
an “equivalent” reference product, or basket schemes were adopted by Germany, the
of equivalent reference products. The stated Netherlands, and New Zealand, beginning
goal of reference pricing is to encourage in the late 1980s and early 1990s. The Dutch
competition by informing consumers about system required that all drugs had to belong
the pricing of substitutable products and to a reference group, while the German sys-
forcing consumers to internalize the incre- tem excluded many drugs from any referenc-
mental costs of a drug over and above that of ing. From a practical standpoint, therefore,
a comparable substitute. the Dutch system imposed some type of
In contrast, price controls explicitly or reference-pricing control on all drugs, while
implicitly limit the prices that manufacturers the Germans selectively imposed it on only
can charge. Given the enormous heteroge- certain drugs. On the other hand, reference
neity in drugs, it is not practical to pass blan- prices within groups were slightly higher
ket price restrictions on all pharmaceuticals. in the Netherlands than in Germany, by
Instead, price controls take forms that are 1–3 percent. Overall, the early incarnations
more easily applied to individual drugs or of the German and Dutch reference-pric-
drug classes. Two common types are direct ing schemes did not substantially reduce
price negotiations, in which governments incentives to launch or to sell pharmaceuti-
use their leverage to bargain prices with drug cals in those countries, because the Dutch
manufacturers, or health technology assess- system offered relatively high reference
ment, in which governments set implicit prices, while the Germany system excluded
price ceilings by mandating minimum many drugs from referencing (Danzon and
cost-effectiveness (CE) thresholds or other Ketcham 2004). On the other hand, New
metrics of price relative to benefit. Zealand set much lower reference prices,
and had relatively broad application of ref-
4.3.1 Reference Pricing
erencing. Perhaps as a result, new drugs
One form of public p rice setting is known were less likely to launch in New Zealand,
as reference pricing. Two subtypes are inter- compared to Germany and the Netherlands
nal reference pricing and external reference (Danzon and Ketcham 2004).
pricing. Internal reference pricing groups External reference pricing is similar in
therapeutically similar drugs together and spirit to internal reference pricing, but uses
uses the prices of similar drugs to set controls prices set in other countries, rather than
or targets within the group. For example, an prices for similar drugs within the referenc-
internal reference-pricing scheme might ing country itself. External reference-pricing
group together all beta blocker drugs as “sim- regimes set o ut-of-pocket costs for a drug
ilar,” and then set as the reference price the based on what the drug or similar drugs
minimum (or median) price charged among cost in a basket of other countries (Espin,
all beta blockers. Consumers purchasing Rovira, and Olry de Labry 2011). The incen-
drugs in this class would pay out-of-pocket tive effects of such schemes can be complex,
the price of the relevant drug, less the refer- because prices and product launches in one
ence price, subject to an out-of-pocket price country may affect a whole host of other
floor equal to zero. countries. Moreover, pricing and launch
Internal reference-pricing schemes vary decisions are often made simultaneously.
along at least two dimensions: the scope of As a general matter, tighter reference-pric-
the reference groups and the mechanism for ing rules that drive down the average
reference price and expand the applicabil- Price controls may also be accomplished
ity of referencing delay drug launches. One through indirect means, most commonly
might reasonably ask why this is true, since through some system for “health technology
even at reference prices, drugs can usually assessment” (HTA). In this case, a government
be sold at a modest economic profit. Indeed, will empower an agency to assess the value of
since costs of production are generally about new health technologies. This assessment of
10–20 percent of peak US
branded-drug value then plays a significant role in the price
prices (Caves, Whinston, and Hurwitz that is offered to manufacturers. Very often,
1991), it requires very aggressive referenc- the influence of the health technology assess-
ing to drive gross margins below zero. This ment on the actual price is somewhat opaque,
may suggest the presence of strategic incen- although there is a correlation such that a
tives. Drug manufacturers wish to signal to poorer assessment leads to a higher likelihood
other countries that adopting tighter refer- of a drug not being purchased by a country, or
ence-pricing schemes will result in failure to being purchased at a lower price.
launch or delays in launch. One common example of health technol-
ogy assessment is the use of CE analysis to
4.3.2 Price Controls
value health technologies. The incremental
An alternative to reference pricing is the cost-effectiveness ratio (ICER) associated

direct or indirect control of pricing in a phar- with a new health technology is equal to the
maceutical market. One example is direct marginal cost of the technology divided by
price negotiation by a large public purchaser the marginal health benefit. Typically, the
against drug manufacturers. Under this marginal health benefit is expressed in terms
scheme, the government or a public agency of additional QALYs, a concept that incorpo-
negotiates with the full buying power of its rates improvements in both the quantity and
citizens behind it. In isolation, direct price quality of life. The ICER is used as a guide
negotiation by a public agency is analogous for pricing. For example, the UK’s National
to the price negotiation that occurs between Institute for Health and Clinical Excellence
private insurers and drug companies in the recommends that every new drug approved
private US market. One expects larger price produce at least one additional Q ALY for
concessions when the manufacturer has less every £30,000 that it costs. C E thresholds
bargaining leverage, and vice versa. From function as a type of implicit price regulation,
a practical standpoint, therefore, prices are where price is tied to the efficacy demon-
expected to be lower for drugs that face more strated in clinical trials (Jena and Philipson
therapeutic competition or are less valuable 2007). It does not seem accidental that C E
to consumers. Sood et al. (2009) show that ratios appear to cluster near the threshold
price negotiation by governments tends to deemed acceptable by the HTA body (Jena
lower pharmaceutical prices, on average, by and Philipson 2007). Manufacturers have
around 20–25 percent. incentives to pick the maximum price that
It is worth pointing out, though, that the still results in acceptable cost-effectiveness.
negotiating behavior of a public agency may A less conventional point arises when one
differ from that of a private insurer. Typically, considers the strategic behavior of manufac-
public agencies face hard, legislated bud- turers facing C E thresholds. In cases where
getary limits. Private insurers may also face CE thresholds are not binding, they may
short-term budgetary limits, but the process actually function as price floors, rather than
that sets budgets is related to p rofit maximi- price ceilings, as manufacturers shift their
zation, rather than legislation. prices upward to meet the threshold (Jena
and Philipson 2009). Regulators may counter suggest that consumers in Hyderabad are
by observing prices set in countries with- less well-informed about quality than those
out such thresholds, but it is clear that the in the Bate, Jin, and Mathur (2011) sample.)
price distortions introduced by such thresh-
4.5 Pricing of Generic Drugs
olds could move in either direction. In this
case, CE thresholds differ fundamentally Much of the economic research on pricing
from factors related to the market power of revolves around branded drugs. However,
buyers. the pricing of generic drugs is attracting
greater interest because generic markets
4.4 Pricing in Emerging Markets
have presented significant puzzles to econo-
Unlike developed countries, emerging mists in recent years.
markets often lack well-developed public or Frank and Salkever (1992) observed that
private insurance arrangements, distribution the entry of generic drugs leads to increases
systems, or quality regulations. On the one in brand prices. They explained this as the
hand, this simplifies the analysis of market consequence of market segmentation.
pricing, since drug companies are, for the Imagine that there are two types of consum-
most part, selling directly to consumers, ers in the marketplace— price insensitive
sometimes without even the intermediation and p rice sensitive. Prior to generic entry,
of a physician. On the other hand, how- branded manufacturers find it optimal to
ever, variation in price within these markets price in a way that captures consumers from
reflects more than just willingness to pay. both these segments. After generic entry,
Differences in drug purity and efficacy are however, it makes more sense to specialize
common. in the p rice-insensitive consumers and to
Bate, Jin, and Mathur (2011) collected allow generics to capture the price-sensitive
samples of prescription drugs from low- and consumers. This behavior may explain why
middle-income countries and found that
branded prices go up after the arrival of
15 percent of the samples were substandard identical generic competitors.
in terms of purity, composition, appearance, This analysis suggests that generic drugs
or at least one other measure of quality. would be priced at razor thin margins to
Notably, they find that substandard drugs retain the most price-sensitive consumers.
have lower average prices than their standard Some have argued that this has contrib-
counterparts, although price was not a per- uted to growing shortages of generic inject-
fect signal of quality in the data. Information able drugs, observed since the mid-2000s.
about quality is thus asymmetric, although Notably, the MMA, which took effect in
not completely so. 2006, reduced reimbursement to health-
Bennett and Yin (2014) find evidence that care providers for such drugs. Specifically,
these quality differentials will tend to dissi- the MMA paid providers a percentage of the
pate if the pharmacy market tends toward drug price, which leads to higher payments
high-volume chain stores, instead of smaller for branded drugs and lower payments for
firms. Analyzing data from Hyderabad, generic drugs. In principle, this could lead
India, they find that chain-store entry manufacturers to abandon the production
improves drug quality and lowers price. of generic drugs. Yurukoglu, Liebman, and
The latter effect suggests the presence of Ridley (2012) argue that this led to shortages
scale economies or other efficiencies that of generics.
may offset the higher willingness to pay for Jacobson, Alpert, and Duarte (2012),
higher quality drugs. (Alternatively, they however, present evidence inconsistent
with this hypothesis. They note that short- too high, and corresponding controversy
ages have been prevalent for non-injectable over whether prices are set too low or too
drugs unaffected by the MMA. They sug- high.
gest that the explanation must involve other Global pharmaceutical pricing also con-
contributing factors, such as episodic man- tains its own normative dimension, distinct
ufacturing problems. Anecdotal evidence from the question of whether pricing is
also suggests that market concentration in dynamically efficient. Consider the decision
the generic industry has led to rising prices. problem of a small country—e.g., Belgium—
The recent episode of daraprim pricing by that is deciding on how to control, negotiate,
Turing Pharmaceuticals is a case in point. or otherwise set pharmaceutical prices. A
Turing Pharmaceuticals acquired the rights small market like Belgium has little influ-
to sell daraprim, which treats toxoplasmosis ence over global innovation outcomes, even
in conjunction with sulfonamides (a class of if it sets a very low price. Thus, there is an
anti-infective drugs). Daraprim is a generic incentive for Belgium to ignore dynamic
drug, but faced no competitors in the toxo- considerations and set as low a price as drug
plasmosis space. After acquiring rights to manufacturers will tolerate. The problem
the drug, Turing rapidly raised the price of of Belgium plays out to different degrees
daraprim by a factor of roughly fifty (Pollack in all markets. No single country internal-
2015). Providers had few good substitutes, in izes the long-run global return to innovation
the absence of alternative suppliers or thera- and, as a result, every individual country will
peutic alternatives. Somewhat unexpectedly, underprice drugs if it has the market power
therefore, lack of competition in the generic to influence prices. These incentives will be
industry might be a significant threat to con- strongest in small markets, and weakest in
sumer welfare. big markets—e.g., the United States—but
they are present in every market (Egan and
4.6 Normative Analysis of Pharmaceutical
Philipson 2013).
Pricing
4.7 Implications for Public Policy on Drug
The efficiency of pharmaceutical pricing
Pricing
turns on both static and dynamic consider-
ations. Obviously, standard marginal cost One of the fi rst-order questions in
pricing paid to the manufacturer is dynam- drug-pricing policy is how to make new,
ically inefficient, because it encourages zero effective medications more affordable for
innovation. Static efficiency can be balanced consumers. Despite all the controversy in
with dynamic efficiency if insured consum- the literature, there seems little doubt that
ers face out-of-pocket costs close to mar- health insurance can help. Providing insur-
ginal costs of production, and manufacturers ance necessarily makes drugs more afford-
simultaneously receive prices in excess of able to consumers and moves prices toward
marginal costs (Lakdawalla and Sood 2013). marginal cost, which is efficient in a static
Indeed, even if pricing is set too high or sense. One can debate whether the result-
too low from a dynamic perspective, static ing increase in utilization results in exces-
efficiency might still obtain. However, this sive innovation or instead moves innovation
observation does not shed light on the ques- up to its efficient level. However, that does
tion of what prices manufacturers should not indict the value of lowering consumer
receive in order to obtain dynamic efficiency. out-of-pocket prices. Static efficiency can be
As discussed earlier, there is great contro- improved while remaining suitably agnostic
versy over whether innovation is too low or about dynamic efficiency issues.
An increasingly important policy ques- It is not hard to understand why drug man-
tion is how to pay for drugs. The model of ufacturers would wish to keep their rebate
quantity-based pricing appears creaky and agreements secret—for one thing, they may
out of date, but regulatory barriers in the US not wish to disclose to some customers that
market stand in the way of rapid innovation they are not getting the biggest rebates in the
in pricing. Allowing more rapid contractual industry. The same is true for payers, who
innovation in pricing thus appears to be a might wish to keep under wraps the very
worthy goal in all markets, and perhaps most generous rebates they might be getting. On
of all in the US market. the other hand, however, the suppression of
Finally, f ree-riding incentives in the global rebate data makes drug prices look higher
pharmaceutical marketplace remain vex- than they really are; this ends up harming the
ing to policy makers. There are not many self-interests of drug manufacturers. Clearly,
obvious policy solutions. In principle, coor- the first consideration appears to outweigh
dination across countries or greater roles the second, although rising pressure for price
for international organizations such as the controls or p rice negotiation may change this
European Union might help. However, it is calculus and lead to release of more rebate
unclear how international alliances can hold data in the future.
together in the face of individual free-riding Another key issue is whether pricing is too
incentives. This problem will pose a major high or too low. This question is intimately
challenge to the pharmaceutical market over linked to the earlier question of whether
the coming years. innovation is too high or too low. Economists
have made strides toward understanding
important parts of this puzzle. We have
Questions
learned how pricing affects the rate at which
True net pricing information is not avail- new drugs are launched. Separately, we have
able to researchers studying the pharma- also learned how some public policies affect
ceutical industry. The rebates paid from pricing. The missing link continues to be the
drug manufacturers to private insurers in cost and value of the marginal drug, as dis-
the United States, and sometimes to public cussed earlier. If we knew those parameters,
payers overseas, are not available publicly, we could answer the question of whether
or even on a confidential basis, for research. increases in price harm or help welfare.
Much of what we know about pharmaceuti-
cal pricing is based on analysis of p
re-rebate
5. Marketing Strategies
pricing data. This is a fairly serious limita-
tion that could significantly distort our view Pharmaceutical marketing is often a key
of the evidence. Indeed, it seems likely that flashpoint for public policy debates sur-
rebates are, in many cases, correlated with rounding the industry. Indeed, the phar-
other variables we care about, like quantity, maceutical industry exhibits relatively high
value, manufacturer market power, payer marketing to sales ratios, compared to other
market power, and other related variables. industries, with marketing expenditures
To date, there have been relatively few rig- at approximately 15–20  percent of sales
orous solutions proposed to this problem. (Berndt et al. 1995). One explanation for this
The literature has proceeded as if stud- is the “experience-good” nature of medicines.
ies on pre-rebate pricing are sufficient for Consumers must try products before they
informing economists about the behavior of know they will work. As a result, there are
the marketplace. substantial returns to inducing c onsumers to
try new drugs, as well as substantial costs of of wasteful or socially suboptimal behavior
initiating a new therapy (Berndt 2002). Both by pharmaceutical firms, who themselves
of these tend to increase marketing expendi- benefit from regulatory or legal patent
tures. Yet, the positive theories tell only part monopolies (Angell 2004). Economists have
of the story. Some have argued that market- pointed out, on the other hand, that market-
ing expenditures are wasteful, provide no ing and innovation investments are comple-
benefit to patients, and inherently trade off mentary, because marketing makes it more
against more productive investments that profitable to sell new products. More gen-
could be made in research and development. erally, economists have made a number of
Economic theory sheds light on both the important contributions to this policy debate
positive and normative aspects of marketing over the appropriate role of marketing
decisions. activity.
5.1 Background on Pharmaceutical 5.2 The Simple Theory of Marketing
Marketing
A good portion of the theoretical literature
The largest component of marketing has on pharmaceutical marketing begins with the
typically been “detailing” to physicians. standard model of monopoly marketing and
This involves the pharmaceutical company its descendants (Schmalensee 1972, Scherer
dispatching sales representatives to visit and Ross 1990, and Dorfman and Steiner
doctors and promote their products. Visits 1954). This model takes the following form:
typically last three to ten minutes, and the
FDA regulates the content. Over the course max   P × Q(P, A)  − MC × Q(P, A)  − A.
P, A
of a detailing visit, sales representatives typi-
cally leave behind product samples, another The choice variables are price ( P)and adver-
significant source of marketing expendi- tising expenditures (A). Quantity is a func-
tures (Berndt et al. 1995). Less significant tion of both. Marginal cost of production is
are ads placed in medical journals, as well MC, which is taken as fixed in this simple
as direct-to-consumer advertising (DTCA), illustration.
which is important for a few, highly selected, In spite of the model’s simplicity, and
high-revenue drugs (Iizuka and Jin 2005,
indeed its lack of realism, a number of use-
2007). ful empirical implications follow that are
DTCA has increased substantially since the particularly relevant to the pharmaceutical
FDA relaxed regulation of it in 1997 (Berndt industry:
2002). Annual spending on DTCA tripled
from 1996 to 2000, when it reached nearly 1. Since allowing advertising raises prof-
$2.5 billion. Even so, this still accounts for itability, it also boosts innovation
no more than 15 percent of total pharmaceu- effort—thus, innovation and advertis-
tical advertising, and remains concentrated ing are complements (Lakdawalla and
among a relatively small number of products. Philipson 2012).
For instance, over the same period of time,
spending on detailing and samples rose by $5 2.
If patent monopolies restrict output,
billion (Rosenthal et al. 2002). advertising can enhance welfare by lim-
The pharmaceutical industry spends iting deadweight loss due to monopoly,
approximately the same amount of money even if the advertising itself conveys no
on marketing as it does on innovation invest- valuable information (Lakdawalla and
ments. This fact is often cited as evidence Philipson 2012).
3.
Advertising levels are higher when However, competitive advertising raises
a unit of sales is more profitable demand for the firm’s own product without
(Lakdawalla, Sood, and Gu 2013). growing overall market demand. Intuitively,
oligopoly models predict that advertising
4.
Holding profitability fixed, advertis- effort is higher when competition is lower.
ing levels are higher when market size With fewer competitors around, individual
is larger (Lakdawalla, Sood, and Gu firms internalize more of the benefits of their
2013). own investments in advertising.
One important special case of the com-
From a normative standpoint, the theory petition result arises in the wake of patent
implies that advertising and innovation are expiration. A firm holding a patent on a drug
complements, contrary to many assertions by internalizes the benefits of all the marketing
some that advertising expenditures “crowd undertaken to support that particular drug.
out” spending on innovation (Angell 2004). After its patent expires, however, benefits
Moreover, the theory implies that advertis- flow to all generic competitors producing
ing, even persuasive advertising that conveys the drug (Lakdawalla and Philipson 2012).
no valuable information, can make consum- This helps account for the widespread pat-
ers better off by limiting monopoly loss. tern that advertising effort falls nearly to zero
The latter implication is less robust than the after patent expiration. It also complicates
others, because it rests more directly on the the effects of patent expiration in the phar-
simplifying assumption that pharmaceutical maceutical industry. On the one hand, pat-
marketing is driven by shifts in consumer ent expirations lower prices by stimulating
demand. This is at odds with findings in the the entry of new competitors. On the other
pricing literature that firms are likely selling hand, patent expirations lower advertising
to an insurer’s break-even curve, or bargain- effort. The first effect boosts utilization,
ing with an insurer, rather than pricing along while the second lowers it. Empirically, pat-
the consumer’s demand curve. ent expiration can either decrease, increase,
Several extensions and relaxations of the or leave unchanged the market quantity of a
simple theory have been developed in the lit- drug (Lakdawalla and Philipson 2012).
erature. First, the model has been extended Dynamic versions of the advertising deci-
into the oligopoly context that more closely sion have also been developed (Bhattacharya
approximates actual competitive conditions and Vogt 2003). Since drugs are experience
in the pharmaceutical industry (Scherer goods, public knowledge about them is valu-
and Ross 1990; Lakdawalla, Sood, and Gu able to their producers. Price reductions
2013).4 The oligopoly extension provides stimulate use and build experiential knowl-
insight into the free-riding problem posed by edge, while marketing builds communicated
advertising. That is, certain forms of adver- knowledge. Since knowledge about a mol-
tising are cooperative, in the sense that they ecule benefits all firms producing the mol-
raise overall market demand, including both ecule, including generic competitors, the
demand for the advertising firm’s product value of knowledge will decrease over the
and demand for its competitors’ products. life cycle of a patent. Therefore, the predic-
tion is for rising prices and falling marketing
over the patent life cycle, both of which seem
4 However, it should be noted that most of the oligopoly
to fit empirical patterns in the pharmaceu-
versions in the literature impose functional form restric-
tions to ensure the simplification that price is set inde- tical industry (Bhattacharya and Vogt 2003,
pendently of advertising. Ellison and Ellison 2007).
Caves, Whinston, and Hurwitz (1991) sug- utilization. Unlike advertising for, say, soda
gest an alternative mechanism for this empir- or sneakers, the chain of causation is rather
ical finding. They study the case in which murky for pharmaceutical advertising. The
advertising is aimed at the subpopulation of use of a particular drug results from a complex
patient with the best available substitutes. and poorly understood joint decision-making
Therefore, to the extent that reductions in problem by patients and physicians. Patients
advertising release these patients to a com- must decide to visit a doctor. The doctor
petitor, the optimal price might rise due to a must decide to diagnose the patient as being
corresponding reduction in the price respon- eligible for a pharmaceutical, and then must
siveness of demand. choose what pharmaceutical to prescribe.
The patient may attempt to influence the
5.3 Empirical Evidence on Pharmaceutical
decision to prescribe anything or the deci-
Marketing
sion to prescribe a specific drug. The patient
It is self-evident that pharmaceutical mar- must follow through by purchasing the drug.
keting must boost utilization of a firm’s own Finally, the patient needs to actually use the
drugs, or profit-maximizing firms would not drug that is purchased.
invest in it. The salient positive questions In several papers, Iizuka and Jin (2005,
are: by how much, and how, exactly? 2007) have argued that DTCA affects initi-
Regarding the question of “how much,” ation of any drug, while direct-to-physician
there exists a range of estimates that appear advertising influences the choice of medica-
to vary across drug classes and types of mar- tion. One straightforward way of interpreting
keting. In the a nti-ulcer market, for example, these findings is to believe that DTCA leads
Berndt et al. (1995) find the sales elasticity patients to seek the advice of a physician
highest for detailing visits (0.553), followed in the first place, while direct-to-physician
by medical journal advertising (0.198), and advertising influences physicians’ patterns of
lowest for D TCA (0.008). More recent stud- prescribing, conditional on a patient visiting
ies of direct-to-consumer marketing find them. Less attention has been devoted to
larger effects. Rosenthal et al. (2003) find the latter part of the drug utilization chain—
that 13–22 percent of prescription drug namely, whether advertising affects the like-
spending growth during the late 1990s was lihood that a patient will fill and ultimately
driven by growth in DTCA. Over their period adhere to a prescription medication.
of analysis, DTCA represented less than 10 Finally, the competitive dynamics associ-
percent of total promotional spending, and ated with pharmaceutical advertising have
thus appeared to be disproportionately pro- been studied. As discussed in the prior sec-
ductive relative to other forms. Indeed, in tion, it has long been understood that the
more recent work, Berndt et al. (2002) argue degree of competition affects incentives to
it is likely that d irect-to-consumer elastici- advertise. A monopolist, for example, has
ties have risen relative to direct-to-physician the strongest incentive to advertise, because
elasticities. They point to increases in the there is no risk that her advertising effort
prevalence of prescription drug insurance, will stimulate demand for competitors.
which increases the base of consumers in Oligopolists, on the other hand, must worry
the marketplace, as well as insurance benefit about this potential for f ree riding. This point
design that is attempting to steer patients to was demonstrated in the wake of Medicare
preferred drugs. Part D’s implementation. The increase in
Economists have also considered the ques- the number of insured consumers made
tion of “how” drug advertising influences advertising more lucrative, because each
additional consumer captured became more causing consumers or physicians to switch

valuable. The implementation of Part D brands, while informative advertising has
boosted advertising by 18 percent overall, the effect of enlarging the market. While
and by 31 percent for drugs in less com- appealing, this assumption has some lim-
petitive classes (Lakdawalla, Sood, and Gu itations. Informative advertising about one
2013). Shapiro (2018) finds similar evidence drug versus its competitors—e.g., Drug A is
for the particular class of antidepressants in shown to produce more adverse side effects
which advertising expands the size of the in heart failure patients than Drug B—can
total market more than it cannibalizes con- cause physicians to switch their prescrib-
sumers from rivals. These findings illustrate ing behavior without necessarily enlarging
that the predictions of the simple theory the total number of patients using drugs.
about competition and advertising appear to As such, informative advertising might also
be confirmed in the pharmaceutical market. induce switching across brands.
This empirical question, however, might
5.4 The Normative Implications of
carry less normative significance than it might
Pharmaceutical Marketing
initially seem. The central empirical ques-
Much debate has revolved around whether tion is whether advertising leads to overuse
pharmaceutical advertising is informative of medication or instead limits its underuse.
or persuasive (Dave 2013). The distinction In a standard monopoly framework, monop-
is relevant from a normative perspective. oly pricing leads to underutilization. The
“Informative” advertising conveys valuable use of advertising mitigates this underuse
information that improves d ecision making. and generates social value as a result. The
It is not hard to see why informative advertis- only exception to this result is if advertis-
ing produces social value. Persuasive advertising triggers so much additional use that the
ing, on the other hand, motivates consumers market quantity ends up “overshooting” the
through means other than informing them. first-best quantity level. From this perspec-
The mechanisms of persuasion stray into the tive, it does not matter whether advertising is
realm of psychology rather than economics, informative, persuasive, or even truthful. If
but a few examples help illustrate the oper- it moves quantity toward the first-best level,
ative distinction. Informative advertising it improves efficiency; if not, it reduces it
might factually describes the health benefits (Lakdawalla and Philipson 2012). The wel-
of a particular drug over and above its com- fare effects of advertising are given by the
petitors, the side effects of one drug versus value of these changes in efficiency, less the
another, or the costs of failing to treat a par- real cost of advertising. Therefore, advertis-
ticular disease. Persuasive advertising might ing is welfare improving if it moves quantity
consist of free pens or thermometers with toward the fi rst-best level, and if the result-
a branded logo that subtly reminds a physi- ing improvement in efficiency outweighs its
cian about a product, or generates a positive cost. Its welfare properties depend on static
emotional affect toward the product. Such efficiency and cost, not on truthfulness.
advertising might influence behavior even if The question of whether advertising
no additional facts are conveyed. moves utilization toward or past first-best
Empirically, it is quite difficult to sepa- utilization is not a simple empirical one, but
rate the effects of persuasive advertising it is a fundamentally different question than
from informative advertising in the phar- asking whether advertising is persuasive or
maceutical industry. Dave (2013) suggests informative. For small-molecule drugs, one
that persuasive advertising has the effect of might argue that utilization does not become
excessive until the marginal benefit of using is clear. Moreover, the high likelihood that
drugs turns negative or nearly so. Due to the innovation levels are not first-best efficient
low marginal cost of these products, even also implies that socially s econd-best optimal
relatively small positive marginal benefits advertising levels may depart from static effi-
may be consistent with efficiency. ciency. To appreciate this point, imagine that
From that perspective, there is some utilization is below the point of static effi-
evidence that increases in DTCA lead to ciency, but innovation levels are still exces-
more adverse drug reactions, even though sive due to subsidized health insurance. In
increases in physician detailing visits appear this setting, increases in advertising promote
to have mixed effects—lowering adverse static efficiency, but compromise dynamic
drug reaction rates for some drugs and rais- efficiency. One would thus need to compute
ing them for others (David, Markowitz, and the net welfare gain from advertising, bal-
Richards-Shubik 2010). This kind of research ancing these offsetting effects. In contrast, if
is well-designed for the specific normative utilization and innovation were both ineffi-
question of whether drug advertising raises ciently low, it would clearly be socially valu-
or lowers static welfare. Naturally, there are able to stimulate advertising investments,
some caveats: most notably, the adverse drug and vice versa.
reporting data, based on the FDA’s Adverse
5.5 Implications for Public Policy Toward
Events Reporting System (AERS), is not
Pharmaceutical Advertising
without flaws. AERS is based on adverse
events temporally associated with the admin- Much ink has been spilled debating the
istration of a drug, and s elf-reported by phy- information content of advertising and its
sicians. Thus, it may contain more reports relative merits versus investments in inno-
for drugs that are “on the mind of” or more vation. Neither of these arguments seems
salient to the physician, and advertising central to the economic welfare analysis of
may play a role in this degree of salience. advertising, which depends primarily on
Nonetheless, the conceptual approach is whether advertising moves us closer to or
promising, in spite of technical limitations of farther from efficient utilization and efficient
the data. innovation. From a purely static perspective,
While we have focused strictly on static policies to stimulate advertising are socially
efficiency, advertising also has normative beneficial if utilization is too low, and vice
implications for dynamic efficiency. As versa. From a purely dynamic perspective,
emphasized earlier, advertising and innova- such policies are beneficial if innovation is
tion are complements. Therefore, restric- too low, and vice versa. From a combined
tions on advertising will tend to dampen perspective, the welfare effects of policies
innovation by reducing p ost-approval profits promoting advertising must compute the
(Lakdawalla and Philipson 2012). If levels of marginal static and dynamic efficiency ben-
innovation are socially suboptimal, this could efits (or costs).
create significant social costs associated with Unless utilization is strictly harming
advertising restrictions. On the other hand, patients on the margin, it seems likely that
if innovation is excessive, then restrictions on utilization is too low for small-molecule
advertising are w elfare improving. We will drugs, which are very cheap to manufacture.
not rehash the debate in the literature over Thus, from a static perspective, policies to
the efficiency of innovation levels, but the stimulate advertising are likely to be w elfare
link between the efficiency of innovation and improving, to the extent that they move utili-
the welfare effects of advertising restrictions zation closer to its efficient level.
Even if one accepts this line of reasoning, harmaceutical innovation, pricing, and
p
however, the dynamic welfare implications marketing. Even relatively simple economic
muddy the waters on their own. There is no models of these phenomena tend to provide
consensus on whether innovation investments a number of testable predictions that have
are too high or too low. If they are too high, been confirmed by the data. Pharmaceutical
then there is a social incentive to limit adver- firms systematically respond to economic
tising that might mitigate or even reverse its incentives when it comes to making invest-
static welfare benefits. In sum, much remains ments in innovation, pricing, and marketing.
unknown about the optimal configuration of This may not seem surprising to economists,
public policy toward advertising. but many influential clinicians, public health
researchers, and policy analysts continue to
be skeptical about the relative importance
Questions
of economic incentives. For example, in a
We have taken some halting steps toward recent and widely cited piece, two highly
understanding and quantifying the wel- influential oncologists argue that cutting
fare impacts of pharmaceutical advertising, pharmaceutical prices would have little effect
but more work is needed. While marketing on innovation, which is motivated primarily
leads to more use, on balance, it is not clear by altruism rather than the prospect of finan-
whether it leads to appropriate use. The cial reward (Kantarjian and Rajkumar 2015).
challenge of studying this question arises It behooves economists to ask searching
from several sources. First, it is difficult to questions about why basic principles of eco-
ascertain the appropriateness of use without nomic theory, which have been tested and
highly detailed clinical information, such as validated in the specific context of pharma-
that found in a patient’s medical chart. The ceuticals, are so profoundly unconvincing
growth of electronic medical records (EMR) to many thoughtful n oneconomists. On the
data, and its availability for research, may one hand, part of the reason may be philo-
slowly help address this problem. To date, sophical and hard to overcome. Clinicians,
however, EMRs remain riddled with miss- for example, may perceive that health-care
ing data that hampers inference. Second, decision making is ultimately about protect-
just as in the case of studying the value of ing the best interests of the patients, rather
new drugs, it is difficult to produce a single than pursuing the self-interest of the deci-
measure of health benefit that applies across sion maker. Such entrenched ideological dif-
a range of drugs. QALYs remain one possi- ferences may, in some cases, be difficult to
ble choice, but they are not available in most influence. On the other hand, however, the
large-scale databases like health insurance inward focus of the economics profession
claims data. Ideally, improvement in the may contribute further to this problem. The
quality of EMR data, and perhaps the inclu- vast majority of seminal empirical papers
sion of QALY data into big EMR databases cited within this manuscript, for example,
or other large-scale patient databases, could have been published by economists, written
address some of the lingering questions for an audience of trained economists, and
about the value of marketing. placed in economics journals. While econ-
omists certainly publish in the clinical and
health services research literature, there
6. Conclusions and Lessons for Policy
are relatively few examples—again draw-
Economists have made major strides ing from the unsystematic sample of papers
in understanding the predictors of cited in this review—of major empirical
studies p ublished outside economics. One References

might reasonably ask whether this is the
Abrams, David S. 2009. “Did TRIPS Spur Innovation?
result of supply or demand. However, at least An Analysis of Patent Duration and Incentives to
one hypothesis is that economists have failed Innovate.” University of Pennsylvania Law Review
to disseminate the empirical power and pre- 157 (6): 1613–47.
Acemoglu, Daron, and Joshua Linn. 2004. “Market
dictive validity of basic economic principles Size in Innovation: Theory and Evidence from the
applied to the pharmaceutical industry. Pharmaceutical Industry.” Quarterly Journal of Eco-
Notwithstanding all the knowledge that nomics 119 (3): 1049–90.
Adams, Christopher P., and Van V. Brantner. 2006.
economists have created about the pharma- “Estimating the Cost of New Drug Development:
ceutical industry, major gaps and challenges Is It Really $802 Million?” Health Affairs 25 (2):
remain. We have stressed the uncertainty 420–28.
Albainy-Jenei, Stephen. 2006. “Claim Construction and
surrounding the normative analysis of the Doctrine of Equivalents in Amgen v Transkary-
innovation investment. The question of otic Therapies, Inc.” Journal of Intellectual Property
whether innovation is too high or too low is Law and Practice 1 (13): 819–21.
Angell, Marcia. 2004. The Truth About the Drug Com-
a first-order—perhaps the first-order—pol- panies: How They Deceive Us and What to Do About
icy question in the economics of the phar- It. New York: Random House.
maceutical industry. Yet, economists have Arevalo, J. Fernando, et al. 2008. “Primary Intravit-
real Bevacizumab for Subfoveal Choroidal Neovas-
not produced a definitive answer. At best, cularization in Age-Related Macular Degeneration:
we have suggestive analyses that point to the Results of the Pan-American Collaborative Retina
low rate of surplus capture by innovators, Study Group at 12 Months Follow-Up.” Retina 28
(10): 1387–94.
and modeled estimates suggesting that the Bach, Peter B. 2014. “Could High Drug Prices
social return to innovation is positive on the Be Bad for Innovation?” https://www.forbes.
margin. While helpful, these findings each com/sites/matthewherper/2014/10/23/
could-high-drug-prices-be-bad-for-innovation/.
require a speculative leap in order to draw Bate, Roger, Ginger Zhe Jin, and Aparna Mathur. 2011.
an inference about the social efficiency of “Does Price Reveal Poor-Quality Drug? Evidence
innovation. Much rides on the answer to this from 17 Countries.” Journal of Health Economics 30
(6): 1150–63.
question. If innovation is indeed excessive, Bennett, Daniel, and Wesley Yin. 2014. “The Market
then public and private payers are wasting for High-Quality Medicine.” National Bureau of
money paying excessively high rewards to Economic Research Working Paper 20091.
Berndt, Ernst R. 2002. “Pharmaceuticals in U.S. Health
firms producing innovations. In contrast, if Care: Determinants of Quantity and Price.” Journal
innovation is too low, lives are being lost pre- of Economic Perspectives 16 (4): 45–66.
maturely due to a lack of medical progress. Berndt, Ernst R., Ashoke Bhattacharjya, David N.
Mishol, Almudena Arcelus, and Thomas Lasky.
What to do about the rising cost of pre- 2002. “An Analysis of the Diffusion of New Antide-
scription drugs is shaping up to be one of pressants: Variety, Quality, and Marketing Efforts.”
the major policy debates of the next decade. Journal of Mental Health Policy and Economics 5 (1):
3–19.
This is an economic question, but one whose Berndt, Ernst R., Linda Bui, David R. Reiley, and Glen
answer must be conveyed to a broad audi- L. Urban. 1995. “Information, Marketing, and Pric-
ence of noneconomists. At the same time, ing in the U.S. Antiulcer Drug Market.” American
Economic Review 85 (2): 100–105.
many critical parameters that bear directly Berndt, Ernst R., Adrian H. B. Gottschalk, and Tomas
on this question remain unknown. Mirroring J. Philipson. 2005. “Assessing the Impacts of the Pre-
the recent history of medical progress itself, scription Drug User Fee Acts (PDUFA) on the FDA
Approval Process.” In Frontiers in Health Policy
economists have achieved much over the Research: Volume 8, edited by David M. Cutler and
past fifty years in the understanding of the Alan M. Garber. Cambridge and London: MIT Press.
pharmaceutical industry, but much remains Berndt, Ernst R., Thomas G. McGuire, and Joseph
P. Newhouse. 2011. “A Primer on the Economics
to be done, and much rides on the success of of Prescription Pharmaceutical Pricing in Health
our research endeavors. Insurance Markets.” National Bureau of Economic
Research Working Paper 16879. Ethical Drug Discovery.” Journal of Economics and
Bhagat, Sanjai, and Ivo Welch. 1995. “Corporate Management Strategy 3 (3): 481–519.
Research & Development Investments International Cockburn, Iain M., and Rebecca M. Henderson. 2001.
Comparisons.” Journal of Accounting and Economics “Scale and Scope in Drug Development: Unpacking
19 (2–3): 443–70. the Advantages of Size in Pharmaceutical Research.”
Bhattacharya, Jayanta, and William B. Vogt. 2003. “A Journal of Health Economics 20 (6): 1033–57.
Simple Model of Pharmaceutical Price Dynamics.” Cohen, Wesley M., Richard R. Nelson, and John P.
Journal of Law and Economics 46 (2): 599–626. Walsh. 2000. “Protecting Their Intellectual Assets:
Bhattacharya, Sudipto, Jacob Glazer, and David E. M. Appropriability Conditions and Why U.S. Manufac-
Sappington. 1992. “Licensing and the Sharing of turing Firms Patent (or Not).” National Bureau of
Knowledge in Research Joint Ventures.” Journal of Economic Research Working Paper 7552.
Economic Theory 56 (1): 43–69. Conti, Rena M., Arielle C. Bernstein, Victoria M. Vil-
Blackstone, Erwin A., and Joseph P. Fuhr. 2013. “The laflor, Richard L. Schilsky, Meredith B. Rosenthal,
Economics of Biosimilars.” American Health and and Peter B. Bach. 2013. “Prevalence of Off-Label
Drug Benefits 6 (8): 469–78. Use and Spending in 2010 among Patent-Protected
Blume-Kohout, Margaret E. 2012. “Does Targeted, Chemotherapies in a Population-Based Cohort of
Disease-Specific Public Research Funding Influence Medical Oncologists.” Journal of Clinical Oncology
Pharmaceutical Innovation?” Journal of Policy Anal- 31 (9): 1134–39.
ysis and Management 31 (3): 641–60. Crew, Michael. 1969. “Coinsurance and the Welfare
Blume-Kohout, Margaret E., and Neeraj Sood. 2013. Economics of Medical Care.” American Economic
“Market Size and Innovation: Effects of Medicare Review 59 (5): 906–08.
Part D on Pharmaceutical Research and Develop- Cudris, Jennifer, and Jaime R. Merchan. 2014. “Immu-
ment.” Journal of Public Economics 97: 327–36. notherapy in Renal Cell Carcinoma.” In Advances in
Boldrin, Michele, and David K. Levine. 2008. “Per- Tumor Immunology and Immunotherapy, edited by
fectly Competitive Innovation.” Journal of Monetary J. D. Rosenblatt et al., 125–47. New York: Springer.
Economics 55 (3): 435–53. Danzon, Patricia M., and Jonathan D. Ketcham. 2004.
Brekke, Kurt R., Astrid L. Grasdal, and Tor Helge Hol- “Reference Pricing of Pharmaceuticals for Medicare:
mas. 2009. “Regulation and Pricing of Pharmaceu- Evidence from Germany, the Netherlands, and New
ticals: Reference Pricing or Price Cap Regulation?” Zealand.” In Frontiers in Health Policy Research,
European Economic Review 53 (2): 170–85. Volume 7, edited by David M. Cutler and Alan M.
Bruen, Brian, and Katherine Young. 2014. “What Garber, 1–54. Cambridge and London: MIT Press.
Drives Spending and Utilization on Medicaid Drug Danzon, Patricia M., and Li-Wei Chao. 2000. “Does
Benefits in States?” Kaiser Family Foundation Issue Regulation Drive Out Competition in Pharmaceuti-
Brief. cal Markets?” Journal of Law and Economics 43 (2):
Budish, Eric, Benjamin N. Roin, and Heidi Wil- 311–58.
liams. 2015. “Do Firms Underinvest in Long-Term Danzon, Patricia M., Sean Nicholson, and Nuno Sousa
Research? Evidence from Cancer Clinical Trials.” Pereira. 2005. “Productivity in Pharmaceutical–Bio-
American Economic Review 105 (7): 2044–85. technology R&D: The Role of Experience and Alli-
Carlson, Josh J., Katharine S. Gries, Kai Yeung, Sean ances.” Journal of Health Economics 24 (2): 317–39.
D. Sullivan, and Louis P. Garrison, Jr. 2014. “Current Dave, Dhaval M. 2013. “Effects of Pharmaceutical Pro-
Status and Trends in Performance-Based Risk-Shar- motion: A Review and Assessment.” National Bureau
ing Arrangements between Healthcare Payers and of Economic Research Working Paper 18830.
Medical Product Manufacturers.” Applied Health David, Guy, Sara Markowitz, and Seth Richards-Shu-
Economics and Health Policy 12 (3): 231–38. bik. 2010. “The Effects of Pharmaceutical Marketing
Caves, Richard E., Michael D. Whinston, and Mark and Promotion on Adverse Drug Events and Regula-
A. Hurwitz. 1991. “Patent Expiration, Entry, and tion.” American Economic Journal: Economic Policy
Competition in the U.S. Pharmaceutical Industry.” 2 (4): 1–25.
In Brookings Papers on Economic Activity: Micro- Davidson, Michael, et al. 1997. “Comparison of One-
economics, 1–48. Washington, DC: Brookings Insti- Year Efficacy and Safety of Atorvastatin versus
tution Press. Lovastatin in Primary Hypercholesterolemia.” Amer-
Chaudhuri, Shubham, Pinelopi K. Goldberg, and Panle ican Journal of Cardiology 79 (11): 1475–81.
Jia. 2006. “Estimating the Effects of Global Patent de Mouzon, Oliver, Pierre Dubois, Fiona M. Scott
Protection in Pharmaceuticals: A Case Study of Morton, and Paul Seabright. 2011. “Market Size and
Quinolones in India.” American Economic Review 96 Pharmaceutical Innovation.” Centre for Economic
(5): 1477–514. Policy Research Discussion Paper 8367.
Chipty, Tasneem, and Christopher M. Snyder. 1999. DiMasi, Joseph A., and Henry G. Grabowski. 2007.
“The Role of Firm Size in Bilateral Bargaining: A “The Cost of Biopharmaceutical R&D: Is Biotech
Study of the Cable Television Industry.” Review of Different?” Managerial and Decision Economics 28
Economics and Statistics 81 (2): 326–40. (4–5): 469–79.
Cockburn, Iain M., and Rebecca M. Henderson. 1994. Dorfman, Robert, and Peter O. Steiner. 1954. “Opti-
“Racing to Invest? The Dynamics of Competition in mal Advertising and Optimal Quality.” American
Economic Review 44 (5): 826–36. Garber, Alan M., Charles I. Jones, and Paul Romer.
Dranove, David, Craig Garthwaite, and Manuel Her- 2006. “Insurance and Incentives for Medical Innova-
mosilla. 2014. “Pharmaceutical Profits and the Social tion.” Forum for Health Economics and Policy 9 (2).
Value of Innovation.” National Bureau of Economic Gaynor, Martin, Deborah Haas-Wilson, and William
Research Working Paper 20212. B. Vogt. 2000. “Are Invisible Hands Good Hands?
Duggan, Mark, and Fiona Scott Morton. 2006. “The Moral Hazard, Competition, and the Second-Best in
Distortionary Effects of Government Procurement: Health Care Markets.” Journal of Political Economy
Evidence from Medicaid Prescription Drug Purchas- 108 (5): 992–1005.
ing.” Quarterly Journal of Economics 121 (1): 1–30. Giaccotto, Carmelo, Rexford E. Santerre, and John
Duggan, Mark, and Fiona Scott Morton. 2010. “The A. Vernon. 2005. “Drug Prices and Research and
Effect of Medicare Part D on Pharmaceutical Prices Development Investment Behavior in the Pharma-
and Utilization.” American Economic Review 100 (1): ceutical Industry.” Journal of Law and Economics 48
590–607. (1): 195–214.
Egan, Mark, and Tomas J. Philipson. 2013. “Interna- Goldman, Dana P., Darius N. Lakdawalla, Jesse D.
tional Health Economics.” National Bureau of Eco- Malkin, John Romley, and Tomas J. Philipson. 2011.
nomic Research Working Paper 19280. “The Benefits from Giving Makers of Conventional
Ekelund, Mats, and Björn Persson. 2003. “Pharmaceu- ‘Small Molecule’ Drugs Longer Exclusivity over
tical Pricing in a Regulated Market.” Review of Eco- Clinical Trial Data.” Health Affairs 30 (1): 84–90.
nomics and Statistics 85 (2): 298–306. Grabowski, David C., et al. 2012. “The Large Social
Ellison, Glenn, and Sara Fisher Ellison. 2007. “Stra- Value Resulting from Use of Statins Warrants Steps
tegic Entry Deterrence and the Behavior of Phar- to Improve Adherence and Broaden Treatment.”
maceutical Incumbents Prior to Patent Expiration.” Health Affairs 31 (10): 2276–85.
National Bureau of Economic Research Working Grabowski, Henry G. 1968. “The Determinants of
Paper 13069. Industrial Research and Development: A Study of
Ellison, Sara Fisher, and Christopher M. Snyder. 2010. the Chemical, Drug, and Petroleum Industries.”
“Countervailing Power in Wholesale Pharmaceuti- Journal of Political Economy 76 (2): 292–306.
cals.” Journal of Industrial Economics 58 (1): 32–53. Grabowski, Henry G., and Nevins D. Baxter. 1973.
Ellison, Sara Fisher, and Catherine Wolfram. 2006. “Rivalry in Industrial Research and Development:
“Coordinating on Lower Prices: Pharmaceutical An Empirical Study.” Journal of Industrial Econom-
Pricing under Political Pressure.” RAND Journal of ics 21 (3): 209–35.
Economics 37 (2): 324–40. Grabowski, Henry G., and John Vernon. 2000. “The
Encaoua, Davd, Dominique Guellec, and Catalina Determinants of Pharmaceutical Research and
Martinez. 2006. “Patent Systems for Encourag- Development Expenditures.” Journal of Evolution-
ing Innovation: Lessons from Economic Analysis.” ary Economics 10 (1–2): 201–15.
Research Policy 35 (9): 1423–40. Grabowski, Henry G., and Y. Richard Wang. 2008. “Do
Espin, Jaime, Joan Rovira, and Antonio Olry de Labry. Faster Food and Drug Administration Drug Reviews
2011. “External Reference Pricing.” Review Series Adversely Affect Patient Safety? An Analysis of the
on Pharmaceutical Pricing Policies and Interventions 1992 Prescription Drug User Fee Act.” Journal of
Working Paper 1. Law and Economics 51 (2): 377–406.
Federal Trade Commission. 2009. “Emerging Health Green, Jerry R., and Suzanne Scotchmer. 1995. “On
Care Issues: Follow-On Biologic Drug Competi- the Division of Profit in Sequential Innovation.”
tion.” Washington, DC: Federal Trade Commission. RAND Journal of Economics 26 (1): 20–33.
Finkelstein, Amy. 2004. “Static and Dynamic Effects of Hall, Bronwyn H., and Josh Lerner. 2011. “The
Health Policy: Evidence from the Vaccine Industry.” Financing of R&D and Innovation.” In Handbook
Quarterly Journal of Economics 119 (2): 527–64. of the Economics of Innovation: Volume 1, edited by
Folland, Sherman, Allen C. Goodman, and Miron Bronwyn H. Hall and Nathan Rosenberg, 609–39.
Stano. 2001. The Economics of Health and Health Amsterdam and Oxford: Elsevier, North-Holland.
Care, Third edition. Upper Saddle River: Prentice Henderson, Rebecca, and Iain Cockburn. 1996. “Scale,
Hall. Scope, and Spillovers: The Determinants of Research
Frank, Richard G., and Joseph P. Newhouse. 2008. Productivity in Drug Discovery.” RAND Journal of
“Should Drug Prices be Negotiated under Part D Economics 27 (1): 32–59.
of Medicare? And If So, How?” Health Affairs 27 Hirth, Richard A., Michael E. Chernew, Edward Miller,
(1): 33–43. A. Mark Fendrick, and William G. Weissert. 2000.
Frank, Richard G., and David S. Salkever. 1992. “Pric- “Willingness to Pay for a Quality-Adjusted Life Year:
ing, Patent Loss and the Market for Pharmaceuti- In Search of a Standard.” Medical Decision Making
cals.” Southern Economic Journal 59 (2): 165–79. 20 (3): 332–42.
Frech, H. E. 1996. Competition and Monopoly in Med- Howard, David H., Peter B. Bach, Ernst R. Berndt,
ical Care. Washington, DC: AEI Press. and Rena M. Conti. 2015. “Pricing in the Market for
Friedman, John N. 2009. “The Incidence of the Medi- Anticancer Drugs.” Journal of Economic Perspectives
care Prescription Drug Benefit: Using Asset Prices 29 (1): 139–62.
to Assess Its Impact on Drug Makers.” Unpublished. Hubbard, R. Glenn. 1998. “Capital-Market
I mperfections and Investment.” Journal of Economic ance.” Journal of Public Economics 93 (3–4): 541–48.
Literature 36 (1): 193–225. Lakdawalla, Darius, and Neeraj Sood. 2013. “Health
Iizuka, Toshiaki, and Ginger Zhe Jin. 2005. “The Effect Insurance as a Two-Part Pricing Contract.” Journal
of Prescription Drug Advertising on Doctor Visits.” of Public Economics 102: 1–12.
Journal of Economics and Management Strategy 14 Lakdawalla, Darius N. Neeraj Sood, and Qian Gu. 2013.
(3): 701–27. “Pharmaceutical Advertising and Medicare Part D.”
Iizuka, Toshiaki, and Ginger Zhe Jin. 2007. “Direct to Journal of Health Economics 32 (6): 1356–67.
Consumer Advertising and Prescription Choice.” Lakdawalla, Darius N, Eric C. Sun, Anupam B. Jena,
Journal of Industrial Economics 55 (4): 771. Carolina M. Reyes, Dana P. Goldman, and Tomas
Jacobson, Mireille, Abby Alpert, and Fabian Duarte. J. Philipson. 2010. “An Economic Evaluation of the
2012. “Prescription Drug Shortages: Recon- War on Cancer.” Journal of Health Economics 29 (3):
sidering the Role of Medicare Payment Poli- 333–46.
cies.” https://www.healthaffairs.org/do/10.1377/ Lakdawalla, Darius N., and Wesley Yin. 2015. “Insur-
hblog20120529.019758/full/. ers’ Negotiating Leverage and the External Effects
Jena, Anupam B., John E. Calfee, Edward C. Mansley, of Medicare Part D.” Review of Economics and Sta-
and Tomas J. Philipson. 2009. “‘Me-Too’ Innovation tistics 97 (2): 314–31.
in Pharmaceutical Markets.” Forum for Health Eco- Lee, Tom, and Louis L. Wilde. 1980. “Market Structure
nomics and Policy 12 (1). and Innovation: A Reformulation.” Quarterly Jour-
Jena, Anupam B., and Tomas J. Philipson. 2007. nal of Economics 94 (2): 429–36.
“Cost-Effectiveness as a Price Control.” Health Lerner, Josh. 1997. “An Empirical Exploration of a
Affairs 26 (3): 696–703. Technology Race.” RAND Journal of Economics 28
Jena, Anupam B., and Tomas J. Philipson. 2009. (2): 228–47.
“Endogenous Cost-Effectiveness Analysis in Health Lichtenberg, Frank R., and Shawn X. Sun. 2007. “The
Care Technology Adoption.” National Bureau of Impact of Medicare Part D on Prescription Drug
Economic Research Working Paper 15032. Use By the Elderly.” Health Affairs 26 (6): 1735–44.
Johnson, Douglas B., Chengwei Peng, and Jeffrey A. Lichtenberg, Frank R., and Joel Waldfogel. 2003.
Sosman. 2015. “Nivolumab in Melanoma: Latest Evi- “Does Misery Love Company? Evidence from Phar-
dence and Clinical Potential.” Therapeutic Advances maceutical Markets Before and After the Orphan
in Medical Oncology 7 (2): 97–106. Drug Act.” National Bureau of Economic Research
Kaiser Family Foundation. 2014. “The Medicare Part Working Paper 9750.
D Prescription Drug Benefit.” Kaiser Family Foun- Loury, Glenn C. 1979. “Market Structure and Inno-
dation Fact Sheet. vation.” Quarterly Journal of Economics 93 (3):
Kanavos, Panos, and Elena Nicod. 2012. “What Is 395–410.
Wrong with Orphan Drug Policies? Suggestions for Malani, Anup, and Tomas J. Philipson. 2011. “Can
Ways Forward.” Value in Health 15 (8): 1182–84. Medical Progress Be Sustained? Implications of the
Kantarjian, Hagop, and S. Vincent Rajkumar. 2015. Link between Development and Output Markets.”
“Why Are Cancer Drugs So Expensive in the United National Bureau of Economic Research Working
States, and What Are the Solutions?” Mayo Clinic Paper 17011.
Proceedings 90 (4): 500–504. Mansfield, Edwin. 1986. “The R&D Tax Credit and
Kremer, Michael, and Christopher M. Snyder. 2006. Other Technology Policy Issues.” American Eco-
“Why Is There No AIDS Vaccine?” https://www.brook- nomic Review 76 (2): 190–94.
ings.edu/research/why-is-there-no-aids-vaccine/ Manski, Charles F. 2009. “Adaptive Partial Drug
Kyle, Margaret K., and Anita M. McGahan. 2012. Approval: A Health Policy Proposal.” The Econo-
“Investments in Pharmaceuticals Before and After mists’ Voice 6 (4).
TRIPS.” Review of Economics and Statistics 94 (4): Martikainen, Jaana, Ismo Kivi, and Ismo Linnosmaa.
1157–72. 2005. “European Prices of Newly Launched Reim-
Lakdawalla, Darius N., Jacquelyn W. Chou, Mark T. bursable Pharmaceuticals—A Pilot Study.” Health
Linthicum, Joanna P. MacEwan, Jie Zhang, and Dana Policy 74 (3): 235–46.
P. Goldman. 2015. “Evaluating Expected Costs and McCutchen, William W., Jr. 1993. “Estimating the
Benefits of Granting Access to New Treatments on Impact of the R&D Tax Credit on Strategic Groups
the Basis of Progression-Free Survival in Non-small- in the Pharmaceutical Industry.” Research Policy 22
cell Lung Cancer.” JAMA Oncology 1 (2): 196–202. (4): 337–51.
Lakdawalla, Darius, et al. 2009. “U.S. Pharmaceutical McGuire, Thomas, Keith Drake, Einer Elhauge, Ray-
Policy in a Global Marketplace.” Health Affairs 28 mond Hartman, and Martha Starr. 2016. “Resolving
(1): 138–50. Reverse-Payment Settlements with the Smoking
Lakdawalla, Darius, and Tomas J. Philipson. 2012. Gun of Stock Price Movements.” Iowa Law Review
“Does Intellectual Property Restrict Output? An 101: 1581–99.
Analysis of Pharmaceutical Markets.” Journal of Law Menell, Peter S., and Suzanne Scotchmer. 2007. “Intel-
and Economics 55 (1): 151–87. lectual Property Law.” In Handbook of Law and Eco-
Lakdawalla, Darius, and Neeraj Sood. 2009. “Innova- nomics: Volume 2, edited by A. Mitchell Polinsky and
tion and the Welfare Effects of Public Drug Insur- Steven Shavell, 1473–570. Amsterdam and Boston:
Elsevier, North-Holland. North-Holland.

Mohr, Penny E., and Sean R. Tunis. 2010. “Access with Scherer, F. M. 2001. “The Link between Gross Profit-
Evidence Development: The US Experience.” Phar- ability and Pharmaceutical R&D Spending.” Health
macoEconomics 28 (2): 153–62. Affairs 20 (5): 216–20.
Mossinghoff, Gerald J. 1999. “Overview of the Hatch– Scherer, F. M., and D. Ross. 1990. Industrial Market
Waxman Act and Its Impact on the Drug Devel- Structure and Economic Performance, Third edition.
opment Process.” Food and Drug Law Journal 54: Boston: Houghton Mifflin.
187–94. Schmalensee, Richard L. 1972. The Economics of
Murphy, Kevin M., and Robert H. Topel. 2006. “The Advertising. Amsterdam: North-Holland.
Value of Health and Longevity.” Journal of Political Schumock, Glen T., et al. 2014. “National Trends in
Economy 114 (5): 871–904. Prescription Drug Expenditures and Projections for
Nordhaus, William D. 1969. “An Economic Theory of 2014.” American Journal of Health-System Phar-
Technological Change.” American Economic Review macy 71 (6): 482–99.
59 (2): 18–28. Scotchmer, Suzanne, and Jerry Green. 1990. “Novelty
Nordhaus, William D. 2004. “Schumpeterian Profits in and Disclosure in Patent Law.” RAND Journal of
the American Economy: Theory and Measurement.” Economics 21 (1): 131–46.
National Bureau of Economic Research Working Scott Morton, Fiona, and Margaret Kyle. 2012. “Mar-
Paper 10433. kets for Pharmaceutical Products.” In Handbook of
Pauly, Mark V. 2009. “Is It Time to Reexamine the Health Economics: Volume 2, edited by Mark V. Pauly,
Patent System’s Role in Spending Growth?” Health Thomas G. McGuire, and Pedro Pita Barros, 763–
Affairs 28 (5): 1466–74. 823. Oxford and Waltham: Elsevier, North-Holland.
Pavcnik, Nina. 2002. “Do Pharmaceutical Prices Shapiro, Bradley. 2018. “Positive Spillovers and Free
Respond to Potential Patient Out-of-Pocket Riding in Advertising of Prescription Pharmaceuti-
Expenses?” RAND Journal of Economics 33 (3): cals: The Case of Antidepressants.” Journal of Politi-
469–87. cal Economy 126 (1): 381–437.
Pekurinen, Markku, and Unto Häkkinen. 2005. “Regu- Sood, Neeraj, Han de Vries, Italo Gutierrez, Darius
lating Pharmaceutical Markets in Finland.” National N. Lakdawalla, and Dana P. Goldman. 2009. “The
Research and Development Centre for Welfare and Effect of Regulation on Pharmaceutical Revenues:
Health Discussion Paper 4. Experience in Nineteen Countries.” Health Affairs
Philipson, Tomas J., Ernst R. Berndt, Adrian H. B. 28 (1): 125–37.
Gottschalk, and Eric Sun. 2008. “Cost–Benefit Anal- Stevens, Warren, Tomas J. Philipson, Yanyu Wu, Con-
ysis of the FDA: The Case of the Prescription Drug nie Chen, and Darius N. Lakdawalla. 2014. “A Cost–
User Fee Acts.” Journal of Public Economics 92 Benefit Analysis of Using Evidence of Effectiveness
(5–6): 1306–25. in Terms of Progression Free Survival in Making
Philipson, Tomas J., and Anupam B. Jena. 2006. “Sur- Reimbursement Decisions on New Cancer Thera-
plus Appropriation from R&D and Health Care Tech- pies.” Forum for Health Economics and Policy 17 (1):
nology Assessment Procedures.” National Bureau of 21–52.
Economic Research Working Paper 12016. Suvarna, Viraj 2010. “Phase IV of Drug Development.”
Pollack, Andrew. 2015. “Drug Goes from $13.50 a Tab- Perspectives in Clinical Research 1 (2): 57–60.
let to $750, Overnight.” The New York Times, Sep- Thamer, Mae, Niall Brennan, and Rafael Semansky.
tember 20. 1998. “A Cross-National Comparison of Orphan
Ridley, David B., Henry G. Grabowski, and Jeffrey Drug Policies: Implications for the U.S. Orphan
L. Moe. 2006. “Developing Drugs for Developing Drug Act.” Journal of Health Politics, Policy, and
Countries.” Health Affairs 25 (2): 313–24. Law 23 (2): 265–90.
Rosenthal, Meredith B., Ernst R. Berndt, Julie M. Toole, Andrew A. 2007. “Does Public Scientific
Donohue, Arnold M. Epstein, and Richard G. Frank. Research Complement Private Investment in
2003. “Demand Effects of Recent Changes in Pre- Research and Development in the Pharmaceutical
scription Drug Promotion.” Forum for Health Eco- Industry?” Journal of Law and Economics 50 (1):
nomics and Policy 6 (1). 81–104.
Rosenthal, Meredith B., Ernst R. Berndt, Julie M. US Department of Commerce. 2004. Pharmaceutical
Donohue, Richard G. Frank, and Arnold M. Epstein. Price Controls in OECD Countries: Implications
2002. “Promotion of Prescription Drugs to Consum- for U.S. Consumers, Pricing, Research and Devel-
ers.” New England Journal of Medicine 346 (7): opment, and Innovation. Washington, DC: U.S.
498–505. Department of Commerce.
Sanjuan, Judit Rius. 2006. “U.S. and E.U. Protection Vernon, John A. 2005. “Examining the Link between
of Pharmaceutical Test Data.” CPTech Discussion Price Regulation and Pharmaceutical R&D Invest-
Paper 1. ment.” Health Economics 14 (1): 1–16.
Scherer, F. M. 2000. “The Pharmaceutical Industry.” Viscusi, W. Kip, and Richard J. Zeckhauser. 2015. “Reg-
In Handbook of Health Economics: Volume 1B, ulating Ambiguous Risks: The Less than Rational
edited by Anthony J. Culyer and Joseph P. New- Regulation of Pharmaceuticals.” Journal of Legal
house, 1297–336. Amsterdam and Boston: Elsevier, Studies 44 (S2): 5387–422.
Ward, Michael R., and David Dranove. 1995. “The Yin, Wesley, A. Basu, J. X. Zhang, A. Rabbani, D. O.
Vertical Chain of Research and Development in the Meltzer, and G. C. Alexander. 2008. “The Effect of
Pharmaceutical Industry.” Economic Inquiry 33 (1): the Medicare Part D Prescription Benefit on Drug
70–87. Utilization and Expenditures.” Annals of Internal
Williams, Heidi L. 2010. “Intellectual Property Medicine 148 (3): 169–77.
Rights and Innovation: Evidence from the Human Yurukoglu, Ali, Eli Liebman, and David B. Ridley.
Genome.” NBER Working Paper 16213. 2012. “The Role of Government Reimbursement
Yin, Wesley. 2008. “Market Incentives and Pharmaceu- in Drug Shortages.” National Bureau of Economic
tical Innovation.” Journal of Health Economics 27 Research Working Paper 17987.
(4): 1060–77.
This article has been cited by:
1. Afschin Gandjour. 2020. A proportional rule for setting reimbursement prices of new drugs and its
mathematical consistency. BMC Health Services Research 20:1. . [Crossref]
2. Difei Geng, Kamal Saggi. 2020. Optimal price regulations in international pharmaceutical markets
with generic competition. Journal of Health Economics 102315. [Crossref]
3. Fanor Balderrama, Lisa J. Schwartz, Christopher J. Longo. 2020. When are Pharmaceuticals Priced
Fairly? An Alternative Risk-Sharing Model for Pharmaceutical Pricing. Health Care Analysis 10. .
[Crossref]
4. David M. Cutler. 2020. Are Pharmaceutical Companies Earning Too Much?. JAMA 323:9, 829.
[Crossref]
5. Rosella Levaggi, Paolo Pertile. 2020. Authors’ Reply to Garattini and Freemantle: “Value-
Based Pricing Alternatives for Personalised Drugs: Implications of Asymmetric Information and
Competition”. Applied Health Economics and Health Policy 56. . [Crossref]
6. Jörg Mahlich, Jörn Sindern, Moritz Suppliet. 2019. Cross-national drug price comparisons with
economic weights in external reference pricing in Germany. Expert Review of Pharmacoeconomics &
Outcomes Research 19:1, 37-43. [Crossref]
7. Pedro Henrique D. Batista, Dennis Byrski, Matthias Lamping, Roberto Romandini. 2019. IP-
Based Incentives Against Antimicrobial Crisis: A European Perspective. IIC - International Review of
Intellectual Property and Competition Law 50:1, 30-76. [Crossref]
8. Colleen Cunningham, Florian Ederer, Song Ma. 2018. Killer Acquisitions. SSRN Electronic Journal
. [Crossref]
9. Juan Pablo Atal, José Ignacio Cuesta, Morten Sæthre. 2018. Quality Regulation and Competition:
Evidence from Pharmaceutical Markets. SSRN Electronic Journal . [Crossref]

Economics of Pharma Industries

Uploaded by

Copyright:

Available Formats

Economics of Pharma Industries

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Economics of Pharma Industries

Uploaded by

Copyright:

Available Formats

Journal of Economic Literature 2018, 56(2), 397–449

The pharmaceutical industry accounts for a substantial chunk of the US economy’s

1. Introduction and many other breakthroughs. At the same

T he pharmaceutical industry represents

Computer and electronic products 334 $77,887 2,951 $26,393

All industries 31–33, 42–82 $365,211 29,327 $12,453

with particular emphasis on their implica- publishing, or professional services. New

competition by generic manufacturers pro- After testing in a laboratory or on animal

Figure 1. Schematic of Pharmaceutical Life Cycle

Source: Mossinghoff 1999. Reprinted with permission of the FDLI.

impacted by several cross-cutting laws. In methods for modifying genetic material of

strong consensus that growth in pharmaceu- Department of Commerce reviewed pricing

had no effect on expenditures in Finland, 3.2.2.1 Background on Alternative Forms

class. Once again, however, it is unclear Even without perfect price-discrimination,

additional consumer captured became more causing consumers or physicians to switch

studies p ublished outside economics. One References

Elsevier, North-Holland. North-Holland.

You might also like

Economics of Pharma Industries

Uploaded by

Copyright:

Available Formats

Economics of Pharma Industries

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Economics of Pharma Industries

Uploaded by

Copyright:

Available Formats

Journal of Economic Literature 2018, 56(2), 397–449

The pharmaceutical industry accounts for a substantial chunk of the US economy’s

1. Introduction and many other breakthroughs. At the same

T he pharmaceutical industry represents

Computer and electronic products 334 $77,887 2,951 $26,393

All industries 31–33, 42–82 $365,211 29,327 $12,453

with particular emphasis on their implica- publishing, or professional services. New

competition by generic manufacturers pro- After testing in a laboratory or on animal

Figure 1. Schematic of Pharmaceutical ­Life Cycle

Source: Mossinghoff 1999. Reprinted with permission of the FDLI.

impacted by several cross-cutting laws. In methods for modifying genetic material of

strong consensus that growth in pharmaceu- Department of Commerce reviewed pricing

had no effect on expenditures in Finland, 3.2.2.1 Background on Alternative Forms

class. Once again, however, it is unclear Even without ­perfect ­price-discrimination,

additional consumer captured became more causing consumers or physicians to switch

studies p ­ ublished outside economics. One References

Elsevier, North-Holland. North-Holland.

You might also like

Figure 1. Schematic of Pharmaceutical Life Cycle

class. Once again, however, it is unclear Even without perfect price-discrimination,

studies p ublished outside economics. One References