Racing To Invest? The Dynamics of Competition in Ethical Drug Discovery
| DOI | http://doi.org/10.1111/j.1430-9134.1994.00481.x |
| Published date | 01 September 1994 |
| Author | Rebecca Henderson,Iain Cockburn |
| Date | 01 September 1994 |
RACING
TO
INVEST?
THE
DYNAMICS
OF
COMPETITION IN
ETHICAL
DRUG
DISCOVERY
IAIN
COCKBURN
REBECCA
HENDERSON
University
of
British
Columbia
and
hBER
Massachusetts
Institute
of
Technology and
NBER
Recent advances
in
the theoretical literature have greatly expanded
our
under-
standing of the forces that shape the competitive dynamics of research
and
development, but
a
paucity of suffzciently detailed empirical data has left these
insights relatively untested. We draw
on
unusually detailed qualitative
and
quantitative infernal data provided at the research program level
by
10
major
pharmaceutical
firms
to explore the usefulness of the modern literature as
a
source of insight into the dynamics of competition in ethical drug discovery.
Our analysis focuses
on
one particularly compelling aspect
of
the litera-
ture: the suggestion that
in
"winner take
all
situations," competition
in
RbD
becomes
a
Prisoner's Dilemma, leading to overinvestment in research. With-
out adequate measures of the social return to innovation, we can say nothing
about whether there is "too much" or "too little" research undertaken
by
the
industry, but our results do not support the suggestion that
RbD
investment
in drug discovery is driven by the "tit-for-tat" or simple reaction function
models hinted at by the institutional literature. First,
R&D
investment
is
only
weakly correlated across firms once common responses to exogenous shocks are
accounted far,
and
second, rivals'
RbD
results are positively correlated with
own research productivity, which we interpret as evidence for extensiae
RbD
spillovers rather
than
the depletion externality implied by "winner take
all"
models.
These results are not, of themselves, sufficient to reject the hypothesis
that investment behavior
in
the industry is driven
by
strategic considerations
This research was funded by the Sloan Foundation and by four pharmaceutical compa-
nies. Their support is gratefully acknowledged. We would also like to express our
appreciation
to
all
of
those firms that contributed data to the study, and to Allan Afuah
and Andrew Hoffman, who provided exceptional research assistance. We thank the
editor and
two
particularly helpful referees, together with participants
in
seminars at
the
NBER,
the
UBC
Summer
I0
Conference, Stanford, MIT,
UCLA,
and Berkeley for
helpful comments and suggestions. The conclusions and opinions expressed
in
this
paper are, however, our own, and any errors or omissions remain entirely our responsi-
bility.
Q
1995
The
Massachusetts
Institute
of
Technology.
lournal
of
Economics
&
Management
Strategy, Volume
3,
Number
3,
Fall
1994,
481-519
482
Journal of Economics
&
Management Strategy
since there is theoretical support for
a
wide variety
of
observable behavior as
equilibrium outcomes of stratesic interaction. Nonetheless, fhey suggest that
the more extreme
forms
of
rent dissipation identified in the literature are
probably poor characterizations of fhe reality
of
competition in pharmaceuti-
cals.
Our
results point both
to
the need to develop theories that incorporate
richer models
of
possible payoff structures, adjustment costs,
and
firm
hetero-
geneity
and
fo the need
to
collect empirical data that
is
comprehensive enough
to enable one
to
test them.
1.
INTRODUCTION
What drives a firm’s decision to invest in research and development?
Traditional industrial organization theory focused on market struc-
ture, demand, technological opportunity, and appropriability as de-
terminants of the intensity and direction of R&D investment, and
several empirical studies have demonstrated the fruitfulness of this
approach (Levin and Reiss, 1988). More recent theoretical work has
stressed strategic interaction among rivals as a primary determinant
of investment decisions, and this approach has spawned a wealth of
interesting and elegant theoretical models that have suggested some
powerful insights into the dynamics
of
competition in R&D. Reinga-
num (1989) provides an excellent summary of this literature. For more
recent work, see, for example, Dasgupta and Maskin (1987), Dixit
(1988), DAspremont and Jacquemin (1988, 1990), Fraja (1993), and
Suzumura (1992).
Despite the potential power of the game theoretic literature as
a
source
of
insight into the dynamics of competition, surprisingly little
empirical research has attempted to build on its insights, and our
understanding of the forces that drive R&D investment decisions in
practice remains sketchy; Khanna (1993), Lerner (1991), Meron and
Caves
(1991),
and Scherer (1992) are notable exceptions. One possible
reason for this paucity of empirical work may be a lack
of
suitably
detailed data. Whereas the early literature took the industry as its
primary unit
of
analysis, more recent work focuses attention on the
individual research project, or
on
the single-project firm,
as
the key
competitive arena. The modern game theoretic literature says little
about the determinants of aggregate investment in
R&D
at the level
of the typical multiproduct, multiproject firm, and firm level data can-
not be used to test its predictions without some heroic assumptions.
In this paper we use unusually detailed data on
R&D
invest-
ments and outcomes gathered at the level of individual research pro-
grams conducted within
10
pharmaceutical firms over a period of more
than
17
years to explore the usefulness of the modern literature as a
Racitzg
to
Invest?
483
source of insight into the dynamics
of
competition in ethical drug
discovery. These highly detailed data are well suited for examining
R&D
investment decisions, since they allow us to draw direct and
meaningful comparisons between investment decisions made by dif-
ferent firms. The pharmaceutical industry also provides a particularly
intriguing setting for this research, since competition in the industry
has often been held up
as
a prime example of the types of strategic
racing behavior predicted by the literature. Moreover, the industry is
a relatively ”pure play” in technological competition. Firms invest
heavily in research and development, and successful research is a key
contributor to commercial success. At the same time, since most firms
invest in a wide range of technologies, we are able to compare compe-
tition across multiple ”markets.” It should be stressed that this paper
examines ethical drug discovery-research intended to identify prom-
ising new drugs-rather than drug development, which
is
the applica-
tion of these discoveries in the clinic. There may be important differ-
ences in the nature of competition in these
two
quite different stages
of pharmaceutical
R&D.
We begin the paper with a review of some of the modern litera-
ture on strategic competition in
II&D.
We argue that in general these
models are very difficult to test. Investment behavior
is
critically de-
pendent on a wide range of factors, including the nature of the payoff
function, the spillover regime, the information structure of the game,
and the extent and nature of asymmetries between players, all
of
which are rather difficult to capture empirically. Moreover, few of the
models in the existing literature can be easily nested.
As
a
first step in the empirical analysis, we therefore focus on
simple correlations in research investment and output across firms at
the program level. We find investment levels are very weakly corre-
lated across firms. This correlation is not robust to the inclusion of
measures of technological opportunity and is consistent with the belief
that levels of investment represent equilibrium responses to common
shocks in demand and opportunity. Firm effects are quite significant,
and investment levels are very highly autocorrelated. We interpret
this as consistent with the hypothesis that adjustment costs and unob-
served firm heterogeneities play
a
significant role in determining in-
vestment patterns.
To
capture out
of
equilibrium behavior, we measure changes in
investment levels using both first differences and a measure of
”news” in investment. In contrast to previous studies that have ex-
plored correlations in first differences
if
investment across firms, we
find no evidence of correlation in either first differences or ”news”
in investment across firms at the program level. We also find no evi-
To continue reading
Request your trial