Sunday, December 10, 2006

Tufts Hangs Tough on Opportunity Cost Analysis

Last week I questioned the $800 million quoted in the press and other blogs as the cost of Pfizer's torcetrapib failure (see torcetrapib: "$800 Million" Failure but Kindler Safe). I doubted this was the real cost and suggested that Pfizer was just quoting the results of a 2001 study by the Tufts Center for the Study of Drug Development.

Mostly, I was upset that journalists would just accept this number at face value as the actual loss due to torcetrapib's failure. I suggested that this was an "estimate" that the industry trots out whenever it wants to argue how expensive it is to develop new drugs.

In my post, I mentioned that the Tufts' estimate was disputed because of the inclusion of "opportunity cost of capital" in the calculation. Apparently, I misunderstood the economics, because the lead author of the study, Joseph A. DiMasi, PhD, Tufts CSDD Director of Economic Analysis, submitted the following tough, no holds barred, comment to this blog in defense of the estimate:
I don't read your blog, so I don't know all that you may have written about R&D costs, but someone forwarded this particular blog entry to me. I won't comment about torcetrapib because I don't know (nor do you) what their actual or projected costs were or what they included (i.e., discovery costs, preclinical development, chemistry, manufacturing and controls R&D throughout the process, all clinical trials for all indications, infrastructure costs for an ongoing concern, interaction with regulatory authorities and preparation of regulatory submissions, etc.). I doubt, given in particular statements that I recall from Pfizer people about what they think average costs are for a more recent period than we analyzed, that they simply took $800 million from our (Tufts) study (which included the costs of failures and what may be called time or financing costs).

I will, however, correct you on what you have written in this blog entry about our study and its methodology. For your information, PhRMA did not, as you wrote, sponsor the study (nor, for that matter, did pharma). You write that the study is disputed. That's true, but, as far as I can see, the ultimate sources of that criticism are those with obvious political agendas and who lack appropriate expertise. I have never seen a criticism of the methodology from a bona fide economist. The paper and its predecessor were published in the most methodologically rigorous journal in the field of health economics. Anonymous referees and the editors of the journal, who are among world's leading health economists, reviewed the methodology.

What you call financing costs were clearly quantified in these papers and distinguished from actual cash outlays. [my emphasis] Economists do not dispute the relevance of the time, or financing, costs. You wrote: "It's like me saying that the cost of my BMW equals the actual $50,000 I spent on it plus the money I didn't earn by failing to invest the $50,000 elsewhere. Well, actually no. It's nothing at all like that. You have confused a consumption good with an investment good. That makes it impossible to present a realistic analogy, but keeping to your context an appropriate analogy would go as follows. It's rather like you paying $50,000 for your BMW in cash today, but dealer won't deliver the car to you for ten years.

If you still doubt this logic, then I have a proposition for you. The next time that you want to buy a BMW send me a $50,000 check instead. I promise to pay you back exactly $50,000 ten years from now. By your logic you should be OK with that. Both possibilities should be equally valuable to you. In fact, I would sweeten the pot and pay you an extra dollar ten years from now so that, by your logic, you would actually be better off by sending me the $50,000 check.
My main takeaway from Dr. DiMasi's comment is that he's trying to scam me out of $50,000! Of course, I refused his offer ("Too bad," he said, "I was hoping you would take me up on my offer.") .

If Dr. DiMasi didn't like my analogy (confusing a "consumption good" with an "investment good"), then he surely won't like this analogy presented by my friend Matthew Holt resident maven on The Health Care Blog:
To illustrate, Lets say me and a friend have drunk 3 beers a day for two years at $10 a day. Let's say instead of drinking for the first three months, I'd invested that money in Google stock instead. Now I spent $6,000 on beer and $1,000 on Google stock. My friend spent $7,000 on beer. But at the end of the 2 years my Google stock made me a profit of roughly $6000. By Tufts' accounting logic my friend spent $13,000 on beer--the $7,0000 he spent and the $6,000 he didn’t make on the Google stock because he spent the first $1000 on beer.

The problem with their logic is that it ignores the expected returns -- his hangover and my expected financial reward. My financial reward is of course analogous to the money pharma makes when its products are successful (which is a hell of a lot more than 1.2bn over 10 years!)
Matthew uses both a consumption good (beer) and Google stock (an investment good) in his analogy! I love his creativity, however, and urge readers to submit other examples illustrating the concept of opportunity cost to help us non-bona fide economists understand.

More Examples
To get you started, here are some more examples I found:
"An example of opportunity cost would be going to the movies. The cost of going to the movie is $9.00 or whatever ridiculous amount of money your movie theater charges. The opportunity cost would be something else you could have done with that time, such as studying." [This from the Teenanalyst.com, a site advising teenagers about investments. I'm sure studying is an opportunity cost uppermost in the minds of teenagers!]

"If a shipwrecked sailor on a desert island is capable of catching 10 fish or harvesting 5 coconuts in one day, then the opportunity cost of producing one coconut is two fish (10 fish / 5 coconuts)." [That's just weird!]

"The opportunity cost of buying a box of Cracklin Oat Bran is one-and-a half boxes of Wheat Chex, if that's your second favorite cereal."
Here's one I just thought of:
Assuming it takes my son only 4 years to complete his undergraduate study at Penn State, I will have spent over $100,000 on tuition, room, board, books, wine-in-a-box, transportation, etc. A college education is clearly an investment good -- there's no payback until 2009 at least, when my son is scheduled to get his degree (a college education isn't worth anything without the degree!). Sure, some of you might say wine-in-a-box shouldn't be considered in the calculation, but then I'd say you are not a bona fide expert (ie, a parent of a college student)!

But wait! I forgot my lost opportunity to invest that $100,000 in the next best thing (whatever that is). I could have made another $100,000 with a better investment. So, I will really be spending $200,000 on my investment in my son's degree. I'll have to factor that in on my next IRS return!

But wait! Suppose my son, God forbid!, quit college in his junior year and never earned his BS degree? Did I actually spend $200,000 on his failed attempt? I mean, could I go around to my friends and relatives and say, it was a $200,000 failure? I don't think so. But, IMHO, that's what Pfizer did with torcetrapib -- it spent something, probably a great deal, but the development of torcetrapib was cut short before approval. The $800 million Tufts estimate doesn't apply to that situation.
However, as Dr. DiMasi says, we don't know what Pfizer actually spent, so it's a moot point.

Anyway, I thought readers (and Dr. DiMasi, if "someone" should happen to forward this blog entry to him) might be interested in what a few other people had to say on this topic. The following is a recent thread from the Pharma Marketing Online Discussion Forum (not all members of this forum have "obvious political agendas" and I doubt there's a bona fide economist among them):
Tufts pegs the cost of developing a biotechnology drug at an even higher $1.2 B in the following release: http://csdd.tufts.edu/NewsEvents/NewsArticle.asp?newsid=69

Do these claims of extremely high drug development costs help pharmas justify the high price tags for their products?

--David Jastrow

------------------
People are considering torcetrapib a major loss. And while it's true that the $1 billion loss is a large onetime loss. You must consider the amount of derivative research data that is now available to Pfizer. I believe Pfizer has a secondary HDL drug in the pipe. How much cross pollination do you think there will be from the data collected from the torcetrapib trial?

--Laurent Laor

------------------
While Tufts includes the projected cost of projects that never make it to market (perhaps those that never get past Phase 1), let's not forget that 55% of their funding comes from "unrestricted" grants and commissioned projects from the pharmaceutical industry.

I personally do not doubt that when you figure in the cost of salaries, benefits, overhead, legal services, attending conferences, publishing study reports, etc., these costs are realistic.

--Michael Altmann

------------------
From a clinical point of view, it is interesting that I can accomplish the same objective with soy, exercise and niacin. Didn't cost $1.2 billion to figure it out, and I don't have the problem of killing off my patients.

--Avery L. Jenkins

------------------
The Tufts studies are certainly very high estimates of drug development costs.
I'd suggest that interested readers check out Public Citizen's critique of the Tufts studies prior to accepting their results. Link here: http://www.citizen.org/pressroom/release.cfm?ID=954

So, yes, the Tufts studies help the drug industry make claims as to why it needs to keep prices high, but the costs of drug development in the Tufts studies are not accurate, thus making the argument for high prices less compelling. And when making a me-too drug or altering the molecule of one's own existing product just enough to market it as new, that certainly costs much less than a billion bucks. Switching a drug to extended release format is also not likely to run anywhere near a billion dollars.

--CP, Clinical Psych Blog

------------------
Does it cost that much to discover and develop innovative products? Yes, if you consider the costs ploughed into all the investigational drugs that fail in each Phase of testing, not to mention those such as torcetrapib that have cost a boatload of money that will never be recouped. BUT, that still doesn't justify the fact that clear profits for Pharma average about 17%-20% when the clear profit margin for most other innovation-based industries is about 12%. The first company to swallow the short-term losses, tell Wall Street to piss-off with the quarterly pressure, and lower their prices just a little will be the winner.

--Siobhán NíBhuachalla, M.P.H.

------------------
The major problem with the Tufts studies (other than their propaganda use) is that the "opportunity cost of capital" is a huge part of the number. I cant think of any reason to count that, as it has a return at the end.

But no problem--the $800m/$1.2 bn number makes the industry feel good. Who cares how it came about?

--Matthew Holt
That's enough on this subject from me -- discuss!

I'm off to Washington, DC, where I am participating in the Healthcare Blogging Summit. Hopefully, I'll have a report on that for Tuesday's posting.

5 comments:

  1. Joseph DiMasi11:24 PM

    I'm glad that you apparently are having fun with this (I'm having a little myself). You have, however, side-stepped addressing directly the logic of my comment. You haven't, for example, told me why my analogy is wrong. Instead, you are effectively throwing up a smokescreen by encouraging people to submit more and more bizarre and flawed analogies.

    Perhaps it was a copying and pasting mishap, but you reprinted all of my comment except the last sentence. That sentence told your readers where to go to find a paper that we wrote that critiques what the critics of the study have claimed ("Assessing Claims About the Cost of New Drug Development: A Critique of the Public Citizen and TB Alliance Reports", http://csdd.tufts.edu/_documents/www/Doc_231_45_735.pdf). This is very relevant, all the more so since you have posted a comment in this entry from someone who points people to the Public Citizen report. Yes, CP, it's fine that people look at the Public Citizen report, but then they should also read our paper pointing out the lapses of logic and serious flaws in handling data that is found in the Public Citizen report.

    John, you are right that I don't like Matthew Holt's analogy and explanation. It's hard to make any sense of it. No, Matthew, the "Tufts accounting logic" (actually, the concept belongs to the fields of economics and finance, not accounting) would not imply that your friend paid $13,000 for his consumption good (beer). And even if we were talking about an investment, why would you focus on a specific asset that you know AFTER-THE-FACT had an enormously high return. You then make the odd comment that our logic "ignores the expected returns." It's not clear to me what you are talking about here, but, in fact, the cost-of-capital that is used as the discount rate to determine time costs in our study is an estimated investor expected return. This is made clear in the paper, which leads me to believe that you have never read the paper (some comments and questions from you, John, suggest the same for you). I don't think that it is too much to ask that one actually read a paper before one criticizes it publicly.

    Now, John, let's get to your analogy. While education is an investment of a sort, of course a father helping his son is not a business investment. You can indeed frame it, though, as an investment with a cost and an expected return. The return to you presumably is primarily a psychic one. That is, there is a benefit to you that comes from just seeing your son do well financially and even from just trying to help your son (unless you don't care about that and you contract with your son to appropriate a part of his future lifetime excess earnings from being a college graduate as opposed to a high school graduate; another, but more realistic, financial benefit is that the investment increases the likelihood that your son will be better able to take care of you in your old age if that were necessary). In this context, then yes, if you could value the expected benefits in monetary terms then you would want to make the investment if the value of the expected benefits equalled or exceeded the out-of-pocket plus time costs of your investment. Your example does nothing to refute what we did in the context of a business investment.

    Your flip comment about the IRS is also irrelevant. You get to deduct your son's educational expenses in the year in which they occur as a matter of social policy (it would also be practically quite difficult to treat them differently). Legislators want to encourage education as an investment in the country's future. You suggest that if the IRS considered time costs then you would be much better off (from being able to deduct, say, $200,000 instead of $100,000). Actually, if the IRS treated your son's educational expenses in the same way as they would a business investment in physical capital then, in theory, they would have to capitalize your expenses and amortize them over your son's entire working life. Believe me, you wouldn't like that.

    Finally, Matthew Holt's last comment in your blog entry is not valid. The fact that time costs here are a large part of the total cost is not a reason to ignore them. They are large because the drug development process is so lenghty. If R&D cash outlays and net returns remain the same over two periods, but the development process becomes lengthier, does Holt think that developers are in no worse a position? Does he not think that there is a cost to the longer development process? Our capitalization process simply adds to cash outlays an estimate of the monetary value of the time component. These estimates can then be compared to the present discounted value at the time of launch of the future stream of net returns. There is no double-counting here, as Holt seems to suggest.

    ReplyDelete
  2. Thank you for your comments, Dr. DiMasi.

    Although this discussion is a bit off topic for a blog devoted to marketing, I hope readers enjoy the dialogue and come away with some useful information.

    Of course, I am having some fun with the examples, be they right or wrong. I really liked the fish vs cocunuts one, didn't you?

    Instead of arguing whether or not opportunity costs should be figured into the estimate, let me ask you this: What part of the $802 million estimate in the 2001 study was attributable to that? You mentioned in a previous comment that this was clearly stated in your paper, which, unfortunatley, I haven't been able to get a copy of.

    ReplyDelete
  3. Joseph DiMasi1:54 AM

    John,

    Approximately half ($399 million) consisted of time costs.

    ReplyDelete
  4. What I find most interesting about all these discussions (and other media articles) is that no one has even mentioned the $1.2 billion Pfizer laid out in cold, hard cash to purchase Esperion Therapeutics, the original developer of torcetrapib (then referred to as ETC-216). Esperion was essentially a one-product company, so that investment has gone pffft.

    So I would posit the actual cost of "development" for this drug to date is actually upwards of $1.2B, since Pfizer was buying a partially-developed drug, and precious little else.

    I'd be interested in hearing Dr. DiMasi's opinion on whether this would qualify as a "development" expense. From an accounting standpoint, you would separate this out to "in-process R&D" and "goodwill". Wonder if Pfizer will have to disclose how much it will have to write off in goodwill. That would be an interesting figure.

    ReplyDelete
  5. Joseph DiMasi5:09 PM

    Brett,

    You raise an interesting point. When a compound is obtained by a company from its acquisition of another company, the full R&D cost specific to that compound should include the earlier R&D expenditures made by the acquired firm. However, the purchase price of the company may be more or less than the by-then sunk R&D costs on the compound (even for a one-compound company). The purchase price will depend on what has been learned about the compound and its potential value in the marketplace.

    That said, I believe that you have your facts wrong. ETC-216 and torcetrapib are different compounds. ETC-216 is a large molecule (recombinant protein) combined with a phospholipid. Torectrapib is a small molecule that had been in development since the early 1990s. Esperion was founded in 1998. Ironically, it was Pharmacia & Upjohn (later Pfizer) that sub-licensed development rights to Esperion. ETC-216 was thought to be a complementary product to torcetrapib. It was an intravenous formulation meant to be used for short-term treatment following a heart attack and other coronary events. Esperion also had a couple of other compounds in mid-clinical development and a number of compounds in early-stage development when Pfizer acquired it in early 2004.

    ReplyDelete