The R&D Returns “Ceiling”

R&D Returns "Ceiling"

The R&D Returns "Ceiling"; Some firms spent far less than $2B per New Medical Entity. Even spending 2-4x, none could create more than 2/year Why?


In late 1990’s it became apparent that the discovery and the launch of new medical entities seemed limited by how we approached R&D, and not by how much firms spent. This visualzation of performance by the top 20 firms was shocking to some… (see below for greater detail)     

This graph made clear that firms varied in the ability to create approvable new medical entities (NME’s).  Some notes on the design: 1)  Lehman counts NME’s instead of drugs with slight benefits over already approved products (“me-too” drugs were so common the FDA formally changed its receptivity to such filings.)  2) The horizontal axis reflects the average R&D budget in dollars spent by the firm over the prior eight years.  We saw this “R&D Ceiling” as provocative, challenging and a learning opportunity.  Best performers could introduce no more than 2NME’s per year.  While some of these spent as much as $2B/NME, some did the same trick for 1/4th of that cost.     

Were the firms spending “only” $0.5B per NME truly more effective?  Did they have some special approach that was so much better? Or was performance just a result of random results – what many folks called serendipity or “luck”?     

Here are the 2000 details – we can compare later years performance in a future post:Firms had VERY variable results - design or luck?     

“R&D is an unmanageable black hole we discard money into…?”  This graph’s suggestion of differing designs for R&D with diverging results fascinated me.  Some argue the pattern is random “good fortune”.  Later analyses I will post here in the future show this is not random luck.  They reveal instead that it is the underlying process or “R&D architecture” that create the “winners” in this universe.    

Later: more of my perspective and the core learning’s from the R&D Ceiling phenomenon. It did not prompt any broad change of approach for another five years for the industry, except for a few visionary leaders. Yes, the value of the NME entering the market should also be a factor when maximizing economic “R&D Returns” but these (operational effectiveness and architecting for R&D performance) were new concepts in the late 1990′s.   
What were your observations on this?   Your current thinking?

This entry was posted in General Interest, In the Pipeline and tagged , , , . Bookmark the permalink.



3 Responses to The R&D Returns “Ceiling”

  1. Rick wobbe says:

    Very thought provoking observations. Here are some hasty thoughts it provoked in me:

    1. It’s not clear what can be concluded from comparisons of R&D spending in a given year and drug launches in the same year. Because drug discovery and development (the stuff the R&D money is spent on) takes 7-12 years, depending on what you choose as your starting point, virtually none of the R&D expenditures in a given year go toward a drug launched that year. It makes more sense to compare drugs launched in a given year with spending from some prior year(s) if you want to learn something about R&D productivity.

    2. Different companies include different things in “R&D”. Some companies include things that other companies would label as marketing (e.g. single topic conferences on their drug, “education” of physicians in order to get them to adopt the product). Lack of a common definition of what goes into the R&D budget bucket makes it tricky, at best, to do apples-to-apples comparisons of productivity between companies.

    3. Ignoring points one and two for a moment, it’s interesting to note that the annual R&D spending of the 5 most productive companies is roughly half the R&D spending of the 5 least productive companies. Hmmm…

    4. I like the idea of focusing on NMEs, it’s closer to the thing of greatest concern: innovation productivity.

    5. Again temporarily setting aside items 1 and 2, it would be interesting to subdivide NMEs in various ways. For example, small molecules vs. biologicals; oral drugs vs. parenterals vs. topicals; by therapeutic area.

    6. I wonder how private companies (e.g. Boehringer Ingelheim) compare with publicly-traded companies (e.g. the ones on your list). Too bad there are so few of the former.

    These are just quick thoughts, but items 1 and 2 loom pretty large in my mind. As far as I know, no one has taken those things into consideration in their analysis (I tried in a project management newsletter article I wrote a couple of years ago and it suggested some interesting things). I’d like to hear your thoughts.


  2. Rick wobbe says:

    Just to clarify my point #1 above, Terry stated that the R&D expenditure figures represented an average of the 8 years’ worth of R&D spending PRIOR to the year in which the NMEs indicated on the Y-axis were launched. Reaching back in time like that is definitely the right thing to do as I said. However, an average that reaches back 8 years basically covers the preclinical and clinical development period, the time in which a drug candidate either sinks or swims. Although it does reflect the period of greatest monetary outlay, it also covers the period of the least innovation, because the activities must follow standardized methods and data analysis to comply with regulatory requirements (i.e. of necessity, “innovation” is minimal during this time in favor of compliance). To examine the productivity of innovation, it’s necessary to reach back a few years earlier in the timeline, to the discovery research phase, which is usually considered about 10 years prior to launch, to capture the time when creativity , rather than conformance, is in demand. Of course, that time is also characterized by tremendous uncertainty (you don’t know which compound will be a drug, you don’t know exactly what are the right experiments to do, the requirements for a drug candidate can shift, etc.), which makes accurately capturing and measuring all the relevant activities and their costs trickier.

    I hope this clears up my prior comment.


  3. I agree the creativity and process that SELECTS an NME happens earlier than seven years prelaunch (excepting some fast-track oncology product, perhaps.) But I would maintain that value added to ultimate launch through innovation comes THROUGHOUT the R&D cycle. For instance, “translational medicine” is more rapidly informing Discovery of the validity of their disease and target models. That technology and process investment is proving value there and across the pipeline’s timeline.

    Our focus is on how to design the best overall R&D process to create higher returns*. This chart merely notes a starting point where we were ten years ago – an imperfect model perhaps, but a starting point. And one that we hope to improve on here. (*financial yes, but this also reflects on how society values those end results.) Such retrospectives and (planned) improvements in rigor MIGHT create additional insights we can act upon. For instance the resulting Tufts meta-analysis revealed a correlation of firms improving “technical” operations (pre-six-sigma-like) to R&D returns. Maybe validating the value of that path prompted others to then follow. There is certainly a lot of six-sigma initiatives and even formal groups within R&D now, ten years later.

    My current work is far closer to today’s urgency on how best to invest NOW in (technologies, innovative transformation or six-sigma-like) ways to improve R&D performance. This retrospective view is intended to create this kind of dialog and perhaps trigger some insights from all contributors. Thanks again to Rick and others that have commented offline.

Leave a Reply

Your email address will not be published. Required fields are marked *