R&D efficiency

Forbes' Matt Herper takes a look at the cost to develop a new drug, and now current estimates put that figure at $1B-$4B.

While the current estimate is newsworthy, folks at places like Tufts have been conducting this exercise for years, and the numbers are always eye-popping (and debatable.)

What makes this particular article interesting is how you can also use the analysis conducted by Herper to compare pharma productivity over the last 15 years. Take a look at the R&D productivity of the top 12 pharmas:

Here's my takeaways:

-There's two tiers of productivity in the analysis: the "productive" cluster (AMGN, NVS, BMS, MRK, ABT, and LLY) all cluster between $3.7B and $4.6B in cost per new drug, while the "less productive" ranged from $5.9B to 11.8B per drug. While half of the companies studied, the "productives" account for 66 of the 135 drugs (49%) these 12 companies introduced in the last 15 years. So you can't say that higher R&D productivity is also a factor of scale - the productive and less-productive companies produced roughly the same number of drugs. 

-The "less-productive" companies tend to be the product of mega-mergers. Each of these companies has done deals to one extent or another, but think of the "biggies" and you're generally thinking of the "less productive" group. Careful, though, when thinking about the time element here - MRK, for example, only did their big SGP acquisition in late 2009. This brings up the question: do mergers depress R&D productivity, or is it mostly companies with declining R&D productivity that have the urge to merge? (My guess: a bit of both, but considering that the 6 most productive companies are generally considered the least involved in the M&A game due to a bias towards internal efforts, it may be a moot point. M&A either distracts from focus, or results in sub-efficient R&D orgs.

(I'm being charitable to NVS, which is a product of a mega-merger (Sandoz and Ciba-Geigy), but that occurred in 1996 - prior to the analysis period. Either NVS did a much better job of integrating R&D, or it takes 15 years to overcome the M&A inefficiencies.)

-I think it would be appropriate to believe that these results also project future R&D efficiency and likely future stock performance.  (e.g. over the next 15 years, AMGN is likely to be much more productive than AZN.) The 15 year period (and $75B in R&D spend) should account for short-term spikes and likely demonstrates which companies have the best R&D people and organizations. I am especially impressed with Novartis (21 products over 15 years) and most disappointed by AstraZeneca (5 products over the same period.) Perhaps this reflects one company choosing easier/harder targets, but I think it more likely reflects capabilities.

-For all of the news and criticism, Pfizer's R&D isn't too bad. The criticism that failures like torcetrapib reflect diminished R&D productivity due to repeated mergers seems misplaced, as Pfizer was almost middle-of-the-pack in R&D efficiency over the last 15 years.

-You might expect that the broadest R&D portfolios would have the smoothest results (success in one area, say cancer, making up for failures in another, say neuroscience.) However, the more productive companies are to me the least broad. Rightly or wrongly, I think of BMS & AMGN biased towards cancer research, while GSK and JNJ are the most diversified. Does this mean that there is R&D value in specialization?

Any other insights to be gleaned from the Forbes analysis?

A couple of caveats to the analysis: 

-The best analysis would weight productivity with resulting product sales. (In other words: you'd accept lower R&D spending efficiency if the output were blockbusters.)

- I can't tell from the Forbes analysis exactly what is included in the figures. I suspect that Roche data includes historical Genentech R&D spending and output. I think DNA has been one of the most efficient AND effective R&D organizations, so I would be very curious to see DNA split out from pre-merger Roche.