Matthew Herper has a really interesting story in Forbes on a new report that attempts to rank biopharma companies by their R&D abilities. Richard Evans of Sector and Sovereign Health (ex-Roche) has ranked companies not on their number of drugs, but on their early-stage innovation. He counts patents, for example, but not the later ones defending existing franchises, and he also looks to see how often these patents are cited by others. As for a company's portfolio, being early into a new therapeutic area counts for a lot more than following someone else, but at the same time, he's also trying to give points for companies that avoid "Not Invented Here" behavior (a tricky balance, I'd think). The full report can be purchased, but the parts that Herper have shared are intriguing.
Ranking the companies, he has (1) Bristol-Myers Squibb, (2) Celgene, (3) Vertex, (4) Gilead, and (5) Allergan. (Note that Allergan is currently being pursued by Valeant, who will, if they buy them, pursue their sworn vow to immediately gut the company's R&D). At the bottom of his table are (18) Novartis, (19) Regeneron, (20) Bayer, (21) Lilly, and (22) Alexion. (Note that Evans himself says that his analysis may be off for companies that have only launched one product in the ten years he's covering). I'm not sure what to make of this, to be honest, and I think what would give a better picture would be if the whole analysis were done again but only with the figures from about fifteen years ago to see if what's being measured really had an effect on the futures of the various companies. That would not be easy, but (as some of Herper's sources also say), without some kind of back-testing, it's hard to say if there's something valuable here.
You can tell that Evans himself almost certainly appreciates this issue from what he has to say about the current state of the industry and the methods used to evaluate it, so the lack of a retrospective analysis is interesting. Here's the sort of thing I mean:
Too often, Evans says, pharmaceutical executives instead use the industry’s low success rates as an argument that success is right around the corner. “A gambler that has lost everything he owned, just because he now has a strong hand doesn’t make him a good gambler,” Evans says. . .
True enough. Time and chance do indeed happeneth to them all, and many are the research organizations who've convinced themselves that they're good when they might just have been lucky. (Bad luck, on the other hand, while not everyone's favorite explanation, is still trotted out a lot more often. I suspect that AstraZeneca, during that bad period they've publicly analyzed, was sure that they were just having a bad run of the dice. After all, I'm sure that some of the higher-ups there thought that they were doing everything right, so what else could it be?)
But there's a particular chart from this report that I want to highlight. This one (in case that caption is too small) plots ten-year annualized net income returns against R&D spending, minus the cost of R&D capital. Everything has been adjusted for taxes and inflation. And that doesn't look too good, does it? These numbers would seem to line up with Bernard Munos' figures showing that industry productivity has been relatively constant, but only by constantly increased spending per successful drug. They also fit with this 2010 analysis from Morgan Stanley, where they warned that the returns on invested capital in pharma were way too high, considering the risks of failure.
So in case you thought, for some reason - food poisoning, concussion - that things had turned around, no such luck, apparently. That brings up this recent paper in Nature Reviews Drug Discovery, though, where several authors from Boston Consulting Group try to make the case that productivity is indeed improving. They're used peak sales as their measure of success, and they also believe that 2008 was the year when R&D spending started to get under control.
Before 2008, the combined effects of declining value outputs and ever-increasing R&D spending drove a rapid decline in R&D productivity, with many analysts questioning whether the industry as a whole would be able to return its cost of capital on R&D spending. . .we have analysed the productivity ratio of aggregate peak sales relative to R&D spending in the preceding 4 years. From a low of 0.12 in 2008, this has more than doubled to 0.29 in 2013. Through multiple engagements with major companies, we have observed that at a relatively steady state of R&D spending across the value chain, a productivity ratio of between 0.25 and 0.35 is required for a drug developer to meet its cost of capital of ~9%. Put simply, a company spending $1 billion annually on R&D needs to generate — on average — new drug approvals with $250–350 million in peak sales every year. . . So, although not approaching the productivity ratios of the late 1990s and early 2000s, the industry moved back towards an acceptable productivity ratio overall in 2013.
I would like to hope that this is correct, but I'm really not sure. This recent improvement doesn't look like much, graphically, compared to the way that things used to be. There's also a real disagreement between these two analyses, which is apparent even though the BCG chart only goes back to 1994. Its take on the mid-1990s looks a lot better than the Evans one, and this is surely due (at least partly) to the peak-sales method of evaluation. Is that a better metric, or not? You got me. One problem with it (as the authors of this paper also admit) is that you have to use peak-sale estimates to arrive at the recent figures. So with that level of fuzz in the numbers, I don't know if their chart shows recent improvement at all (as they claim), or how much.
But even the BCG method would say that the industry has not been meeting its cost-of-capital needs for the last ten years or so, which is clearly not how you want to run things. If they're right, and the crawl out of the swamp has begun, then good. But I don't know why we should have managed to do that since 2008; I don't think all that much has changed. My fear is that their numbers show an improvement because of R&D cuts, in which case, we're likely going to pay for those in a few years with a smaller number of approved drugs - because, again, I don't think anyone's found any new formula to spend the money more wisely. We shall see.