New Challenges in Alliance Portfolio Management - Innovation and Marketing in the Pharmaceutical Industry - page 155

Illustration: Technology Diversity Versus Product Market Diversity
To substantiate that different forms of diversity are empirically distinguishable,
I briefl y refl ect on a pharmaceutical study into the performance consequences
of product market diversity and technology diversity (Wuyts et al. 2010 ). The
study is limited to diversity within fi rm boundaries and corroborates that tech-
nology diversity differs empirically from closely related forms of diversity.
Product market diversity is likely a profi table strategy. A pharmaceutical
fi rm with a strong reputation in a given therapeutic class, access to distribu-
tion channels, and accumulated knowledge on testing and approval proce-
dures may carry over these benefi ts to any new product market it enters giving
it an edge over less experienced competitors. Also technology diversity is
likely a profi table strategy as it is associated with experimentation and recom-
bination, better match with customer requirements, and more valuable inno-
vations, all of which increase performance (Argyres 1996 ; Clark and Fujimoto
1991 ; Kodama 1992 ). Because of internal resource commitments, fi rms likely
experience decreasing returns to further internal diversifi cation: we expect
log-linear effects of diversity on profi tability.
To test whether both types of diversity exert separate effects on perfor-
mance, we analyzed 29 pharmaceutical fi rms over 2 decades (1982-2001).
The fi rms produced around 80 % of the drugs listed in the Food and Drug
Administration (FDA) database and close to 75 % of the patents owned by all
pharmaceutical fi rms. We operationalize “product markets” as therapeutic
classes. For each sample fi rm, we collect information on its approved new
drugs (NDAs) from the U.S. FDA. Following Ellison et al. ( 1997 ), we use the
Uniform System of Classifi cation (USC) of Intercontinental Medical Statistics
(IMS) which categorizes drugs into therapeutic classes on the basis of 5-digit
codes. The degree of therapeutic substitutability is much greater within than
across these therapeutic classes. For example, the anti-infective drugs Kefl ex
and Ceclor (Eli Lilly), Duricef and Ultracef (Bristol Myers Squib), and
Velosef and Anspor (SmithKline Beecham) share the same USC code (15310)
and are close therapeutic substitutes (Ellison et al. 1997 ). On the basis of ther-
apeutic classes, we construct an entropy measure of product market diversity
(Palepu 1985 ). Our measure of technology diversity is based on patent classes,
which capture the notion of distinct “technological areas” (Lerner 1994 ;
Moser and Nicholas 2004 ). Much of the pharmaceutical industry's intellec-
tual property is captured in patents. Patent information and classifi cations are
obtained from the US Patent and Trademark Offi ce (USPTO) and Community
of Science (COS) databases for the sample fi rms. Each patent is assigned a
9-digit US classifi cation code by USPTO, which corresponds to a position
within the hierarchical technology classifi cation system. This system has been
used in a number of papers to study issues related to innovation and technology
(e.g., Lerner 1994 ; Moser and Nicholas 2004 ; Paruchuri 2010 ). Analogous to
(continued)