Modeling the Effects of Promotional Efforts on Aggregate Pharmaceutical Demand: What We Know and Challenges for the Future - Innovation and Marketing in the Pharmaceutical Industry - page 609

effective when patients are not yet fully educated about the disease treated by the
drug whereas brand advertisements are more effective once patients are educated.
Finally, detailing is potentially most effective in increasing category sales, however,
competitive effects reduce the effectiveness to levels not statistically different from
zero. These results are of great value to managers responsible for the allocation of
marketing budgets over brands and over time as well as for policy makers needing
to understand the complex dynamics when developing health care policy.
Given their importance to managers and policy makers, as well as their contribu-
tion to model fi t, temporal differences in pharmaceutical marketing effectiveness
are not to be ignored. First, modelers need to ensure that model parameters of brands
that are introduced at different stages of the category life cycle are not pooled.
Second (and particularly when focusing on new brands, or brands that are close to
patent expiration), it is important to specify models with time-varying parameters.
Several methods for incorporating time-varying parameters are available to research-
ers. Particularly well-suited are the state space and related dynamic linear models
used by Kolsarici and Vakratsas ( 2010 ), Osinga et al. ( 2010 ), and Keller and Pauwels
( 2009 ). These approaches are increasingly popular in marketing, due to their fl exi-
bility and suitability for analyzing long-term phenomena (Leefl ang et al. 2009 ).
Other suitable approaches for estimating time-varying parameters include the
penalized splines method (e.g., Stremersch and Lemmens 2009 ) or moving- or
rolling-window analysis (e.g., Pauwels and Hanssens 2007 ). Third, modelers need
to be careful when interpreting the infl uence of pharmaceutical marketing on cate-
gory sales. Fischer and Albers ( 2010 ) show that it pays to perform analyses at the
brand level, even when being interested in category level effects.

20.5 Conclusions and a Research Agenda

This chapter reviewed studies on the effectiveness of pharmaceutical marketing on
aggregate demand. We fi rst presented an overview of earlier research in this fi eld
and subsequently considered two challenges that are particularly relevant for the
pharmaceutical market: how marketing variables affect the diffusion pattern of
newly introduced pharmaceutical innovations and how dynamics infl uence pharma-
ceutical marketing effectiveness.
From the overview in Sect. 20.2 we conclude that marketing efforts generally
have positive, but moderate effects on pharmaceutical demand. We also fi nd that the
response to pharmaceutical marketing is heterogeneous, both at the product category
level and at the brand level. The wide range of marketing instruments, products, cat-
egories, and markets that these studies cover suggest moderators that might explain
the heterogeneity in the observed effects. The meta-analysis of Kremer et al. ( 2008 )
confi rms that the marketing instruments differ in their effectiveness. For the brand
level studies detailing is found to be the most effective instrument, followed by jour-
nal advertising, meeting expenditures and direct mail. DTCA appears to be the least
effective at the brand level. The effi ciency ranking of instruments is less clear at the