In an article titled "The Make-Believe Billion - How drug companies exaggerate research costs to justify absurd profits" in Slate the author writes about a new study about the costs of drug development.

The common cost for the development of a new drug that I have heard several times, and that is also mentioned in the article as a comparison, is about 1 billion US dollars. The new study now seems to claim that the real cost is only 55 million US dollars.

The statistic Big Pharma typically cites [...] is that the cost of bringing a new drug to market is about $1 billion. Now a new study indicates the cost is more like, um, $55 million.

Is "Big Pharma" exaggerating the development costs for new drugs or is this new study underestimating the true costs enormously?

  • 1
    A large portion of the development of new drugs comes from state subsidies, so the cost to the company that makes it can be viewed in two different ways. The pirate party Sweden (which is highly biased in this question) claims 85% of the development cost is currently paid by the state (piratpartiet.se/politik/patent), and I've seen one graph (though I can't recall where) that while it's lower in the US, the state still funds a majority of it.
    – Kit Sunde
    May 12, 2011 at 19:13
  • 7
    I'm going to try to phrase this gently. If you (a random researcher or some bloody government which wants so much to lower cost of drugs) are so 100% sure that R&D costs of a new drug are so cheap, what's stopping you from staring a brand new "Small Pharma" company, do that research and undersell "Big Pharma"? The fact that this has not happened yet, to a logical and skeptical observer (who has a first clue about how capital markets work), would irrefutably mean that this claim is 100% untrue.
    – user5341
    May 12, 2011 at 20:32
  • 14
    There also needs to be some means of accounting for all of the potential drugs that do not make it to market. Some of those get all the way to late stage trials before it is clear that they will never be able to be sold. If you assume the equivalent of only 17 duds for each one that can be marketed, then the number 55 million is possibly not in conflict with the number 1 billion if one is measuring the cost strictly attributed to a single drug, vs. the other being total R&D divided by the number of drugs that make it.
    – RBerteig
    May 12, 2011 at 22:32
  • 2
    @Konrad - do you have a cite for the last statement? (lots of different University labs do most of the non-marketable fundamental research)?
    – user5341
    May 13, 2011 at 13:31
  • 2
    @DVK - You do require marketing, we just call it "Lobbying" at that point, since you are only marketing to politicians. May 13, 2011 at 17:25

1 Answer 1


The title question isn't really answerable. But there are certainly other sources which bear on that estimate. I'm going to gloss over the different estimates here for a moment. The article claimed the W&L estimate was $55M, one of the authors (I'll quote below) claimed $59M, and I've heard $43M and $53M in reference to this article... let's just say "approximately $50M".

To start with the obvious one, the study was primarily a counter to DiMassi's study that concluded the cost was around $800M per drug.

There are some important considerations to make on this study. Roughly half of this $800M was "cost of capital" associated with keeping the pipeline open for an average of 96months/8years (the average length of time it takes to get a drug through clinical trials). For those of us who don't work with millions of dollars spent over decades, I don't think it's unfair to characterize the cost in the DiMassi study as $400million. Another criticism of the DiMassi study is that the source data was somewhat opaque. A third criticism (made in the L&W paper) is that the companies take excessive tax credits on R&D, and that drives their real costs down from what DiMassi claims.

Although I feel the $800M figure may be overstating things a bit, this newer study also has some severe problems. The most obvious issue is the cost of a phase III trial. Phase III trials usually have between 1,000 and 3,000 subjects, with an average cost of $26,000 per patient(“Clinical Operations: Accelerating Trials, Allocating Resources and Measuring Performance”, 2006). If you assumed no Phase III trial ever failed, that would still almost eat up the entire W&L cost estimate, just for a a phase III trial of 2,000 patients. Given that the failure rate is approximately 50%, I don't see how the ~$50M figure is defensible. You still have to accommodate the costs of pre-clinical development, Phase I and II trials and manufacturing development.

Some additional problems with the estimates come out when the original authors were questioned on it. Rebecca Warburton said in the comments section of that Slate article:

Our estimate of $59 million is the median development (the “D” in R&D) cost per average drug, not just NMEs (new chemicals) and does not include basic research costs, for which there is no reasonable estimate available.

This is in spite of the fact that they referred to "R&D" throughout the paper, and actually did factor in 'basic research' costs to their estimate, while claiming that companies hardly do any. (I'll deal with that at the end).

Rebecca also claims that they did take failed drugs into account.

The discussions with the authors attached to that article are worth reading. Donald Light seems to be saying that he took DiMassi's estimate, and then subtracted everything he thought was unreasonable... rather than approaching the task at hand and estimating costs/discounts/etc.

For a direct confrontation between DiMassi and Donald Light, you can visit the abstract comment section of the L&W paper. One important point to make (from that comment section) is that the R&D costs of a particular drug have no particular bearing on its price. Drug development is a sunk cost, after it is developed, the people who are buying it are paying for the development of the next generation of drugs. This is not a complex idea, and one that both DiMassi and Light understand (although there appears to have been a bit of a vocabulary misunderstanding early on)

The claim was made both in the L&W paper and in the comments section of this question that "most drug discovery is not done by companies". There actually is some data out there in the form of this study. The sources of NMEs that eventually become drugs are:

  • 58% from pharmaceutical companies. (big pharma)
  • 18% from biotech companies.. (startups, VC-funded ventures)
  • 16% from universities, transferred to biotech.
  • 8% from universities, transferred to pharma.

The paper is quite worth reading for other breakdowns (universities are responsible for starting roughly 1/3rd of the 'innovative' drugs that make it to market, which is a significant increase).

TLDR: The $55M estimate is almost certainly too low. But the previous estimate of $800M is almost certainly too high. Nobody has the data we would like to have on this topic.

(Note that my lack of rep here forbids me from using too many links, a somewhat ridiculous problem on a site that could basically be named "citationneeded.com")

  • 3
    The cost of capital is not an outrageous thing to include, if it can be quantified reasonably. Another issue that must be taken into account is that the patent runs for 17 years, but that clock starts ticking many years before sale of the drug is legal. I've heard the claim that some companies have ended up with as few as three to five years of patent protection. When the patent expires, a generic version can be approved for sale typically at a substantial discount from the original.
    – RBerteig
    May 23, 2011 at 7:59
  • 1
    @RBertig: L&W argue that you can either consider research cost a business expense (which you then deduct from taxes) or a long-term investment of capital. Since companies do claim the costs as tax-deductible business expenses, one shouldn't use cost of capital when calculating the cost of a drug.
    – Jonas
    May 23, 2011 at 13:16
  • @Jonas: Do they take the cost as a tax-deductible business expense? or do they take tax credit under the myriad of R&D-focused tax breaks?
    – user73917
    May 24, 2011 at 5:38
  • 1
    How an expensive gets treated for tax purposes is only one factor affecting its actual cost. Tax law may lump all sorts of things together for the sake of simplicity. Not accounting for a cost doesn't make it disappear in the real world.
    – KonradG
    Sep 20, 2011 at 17:23
  • @Jonas I'm not super well-versed in all of this, but I suspect that dichotomy conflates two kinds of capital: capital in the form of research output and capital in the form of equipment (etc.) used to conduct research. The "research is not both expense and investment" dichotomy argues that research output isn't capital investment. OTOH, I think the "cost of capital" that DiMassi uses is the second kind. I.e., researching a drug makes use of lab space, equipment, etc. that costs $$$$$$. You need to divide that $$$$$$ by the ~ # of drugs it's used for, and include it in the cost of each.
    – user37595
    May 9, 2017 at 23:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .