On NPV, DCF and IRR(elevance)?

By: Steven Holtzman, President and CEO, Decibel Therapeutics Inc.

Confronted with the business imperative and opportunity to in-license or acquire a late-stage clinical product candidate, biopharmaceutical leadership teams immediately turn to financial analysis tools to evaluate the asset.

No doubt NPV, DCF and other financial analyses have a role in defining the bounds of rationality for a purchase or sale price. But this type of financial analysis fails to account for different buyers’ ability to withstand a failure, as well as how badly they need the asset.

Consider this: You and I go to our corner convenience store every Friday morning and purchase $100 worth of lottery tickets. The projected DCF of our purchases is exactly the same; equally so with the NPV.

But I make $30,000 per year. You are a private equity maven who makes $50 million per year. We all know intuitively that what I am doing is irrational and value destructive, while in your case there is no issue with your behavior. In fact, it is arguably irrational for you not to buy those tickets.

What this suggests to me is that NPV/DCF analysis, in and of itself, is an insufficient basis for making the decision whether or not to make a purchase — at least in situations characterized by high risk, discordant relative downsides to the buyers and a bimodal outcome.

That fairly well describes what a biotech CEO is facing when contemplating a large acquisition or in-license of late-stage clinical assets. Like a lottery ticket, the outcome of the purchase is fundamentally bimodal: you win or you lose (and, for simplicity, let’s assume you either lose it all or win big).

Hence, the NPV/DCF-based value, incorporating as it does a probability of success (POS), is the one value the asset will never be worth. Indeed, POS is essentially irrelevant (IRR) to the valuation of a late-stage product candidate.

A far better approach is to rely on the courage of your best scientists’ or clinicians’ convictions, and ask them: do you think this puppy will make it, yes or no? If “It’s 50/50” is the best they can do, forget the asset. Unless your key people are ready to stake their reputations on the line — fully recognizing that drug discovery and development is a brutally tough game defined by more failure than success — anything you are likely to pay for it is too much.

Once the scientists/clinicians have aligned behind an asset, three analyses are relevant to determining what you should be willing to pay.

  • Relative Upside (RU): If we are right that the asset will succeed (100% POS), what will it be worth and how important or material a difference will that make to our future?
  • Relative Downside (RD): If we are wrong about the asset (0% POS) and have to write off the purchase price, how important or material a difference will that make to our future?
  • What If We Don’t (WIWD): If we don’t make this move or one like it now or soon, will we be in trouble? If yes, and if we pass on this one, how many more, if any, are there out there?

Two paradigm cases

A few years back, Gilead Sciences Inc. paid about $ 11 billion to acquire Pharmasset Inc. for its late-stage HCV product candidate. Last year Celgene Corp. paid about $7 billion to acquire Receptos Inc. for a late-stage multiple sclerosis (MS) and inflammatory bowel disease (IBD) product candidate. In neither case can the purchase be explained solely in terms of an NPV/DCF analysis.

For Gilead, the RU was obvious: a product that sells in excess of $10 billion a year. The RD was quite large at the time of the purchase; with not quite $30 billion in market cap and about $5.5 billion in cash, an $11 billion purchase was a huge gulp for Gilead and would have been fairly disastrous if it had been wrong.

Nevertheless, the WIWD overwhelmed the RD: Absent securing its predominant position in the antiviral space, Gilead’s future growth would be placed seriously in doubt. And Gilead judged that the Pharmasset asset was essentially the only thing out there that would allow it to retain its position.

In the case of Celgene, the RU was judged to be peak sales in the $4-4.5 billion range. The RD was relatively small. With a market cap above $75 billion and a huge cash flow, Celgene could afford to be wrong about and write off the purchase price without materially damaging the company.

Finally, what I think was really compelling for Celgene was the WIWD. Facing the likelihood of flattening sales of cancer drug Revlimid lenalidomide later this decade, and with the full impact of its large portfolio of investments in early stage discovery not likely to occur before the early to mid-2020s, Celgene was confronting a challenging top-line growth profile in the 2018-23 period. The Receptos asset, if successful, could fill that trough in a way in which very few, if any, other available late-stage assets could. And, arguably, there were not any others available.

Moral of the story

In Book I of his Nichomachean Ethics, Aristotle distinguishes between the goals of different types of human inquiry. In math and logic, we seek universal truths; our goal is “sophia,” or theoretical wisdom (personified by Plato). In matters of human action, we seek knowledge of the particular; our goal is “phronesis,” or practical wisdom (personified by Pericles). Aristotle cautions us that “It is the mark of an educated mind to expect that amount of exactness in each kind which the nature of the particular subject admits.” [ Book I. iii. 4]

For a certain cast of humanity, there exists a deeply seated compulsion to try to render the fundamentally subjective as objective as mathematics and logic, to escape the need and responsibility for human judgment. Financial analysis seems to promise to address this need. However, we are tricked into confusing the extraordinary precision of financial analysis with accuracy.

In his Philosophical Investigations, Wittgenstein, echoing Aristotle, cautions against this desire to escape the “messiness” of human life. Writing of the desire to reduce ordinary discourse to logic, a desire very analogous to seeking to reduce business decisions to financial analysis alone, he writes, “the more precisely we examine ordinary language, the sharper becomes the conflict between it and our requirement.  (For the crystalline purity of logic was, of course, not a result of investigation; it was a requirement.)  The conflict becomes intolerable; the requirement is now in danger of becoming empty.  –We have got on to the slippery ice where there is no friction and so in a certain sense the conditions are ideal, but also, just because of that, we are unable to walk.  We want to walk: so we need friction.  Back to the rough ground!” [Section 107]

The biotechnology industry began with renegade academic scientists, entrepreneurs and a handful of venture capitalists who invested based on their belief in people. What banded these pioneers together was a sense that biotechnology was not merely a business or a job, it was a mission and vocation.

Forty years on, biotechnology has transmogrified into something much more respectable. An Ivy League or Stanford M.D./Ph.D., plus an M.B.A. from Harvard, Wharton or Stanford, are rapidly becoming the price of entry into the field.

With this, the analytic tools of the trade, paradigmatically NPV and DCF analyses, have assumed a paramount place in business decision making.

But what has equally occurred, at a deeper, more personal and individual level, is the replacement of the goal of fulfilling one’s vocation with the primacy of building and preserving one’s career.

As David Brooks put it in the New York Times, “Let’s start with a refresher on the difference between a vocation and a career. A career is something you choose; a vocation is something you are called to. A career is a job you do as long as the benefits outweigh the costs; a vocation involves falling in love with something, having a conviction about it and making it part of your personal identity. A vocation involves promises to some ideal, it reveals itself in a sense of enjoyment as you undertake its tasks and it can’t be easily quit when setbacks and humiliations occur. As others have noted, it involves a double negative — you can’t not do this thing. [August 24, 2016]

To build a career, it is of paramount importance not to make, or be seen to have made, a mistake. Fulfilling one’s vocation, on the other hand, requires putting one’s self at risk of being publicly wrong and taking responsibility for your error. It is in those moments of critical decision making that the soul is revealed.

At least for now, building a great biotechnology company will remain not just a matter of correctly applying business analytics; it will require leadership marked by a sense of mission and vocation, character, judgment and guts.

Author’s disclaimer: I am terrible at arithmetic, find financial analysis largely mystifying, and am neither an M.D./Ph.D. nor an M.B.A. My sole qualification for undertaking this writing is that I have been hanging in and around the biotechnology industry for 30+ years. Wisdom being the product of experience combined with intelligence, I claim only the experience, not the intelligence, much less the wisdom.

Comments

Anonymous says:

Typically, there is much less risk, in the short term, to saying no vs saying yes. The most important learning for me is the reluctance most people have to making an affirmative decision where risk of being wrong is near term and measurable. Decision are made first based on what it means to the decision maker and second on the impact to patients and shareholders.

Anonymous says:

I talked over the summer with some folks at a local pharmaceutical company who expressed something similar to “The Courage of Your Conviction,” if I am reading correctly — firms spend too much time on NPV projections when the real question is, from a scientific perspective, “is this drug a dud or not?”
Are you thinking of doing something with this? It seems like something that Nature Reviews Drug Discovery might be interested in?

Anonymous says:

Complete agreement here. Having participated in or observed countless M&A valuation exercises, it is invariably the result that the probability adjusted DCF of a publicly traded biotech with a development stage asset is equal to its current market cap. Not surprisingly, the models of investment analysts generally come out to the same place, but being analysts trying to make the case for investment they will also generally show the value of a success scenario to justify a target price in excess of the current market cap.

One might ask, why is it that these marvelous DCF analyses almost always recapitulate the current market cap of the company? My argument is that the market is efficient, and the current price of a company represents the then existing equilibrium between buyers and holders (i.e. believers) and sellers (disbelievers). In the absence of inside knowledge or unusual scientific insight, investors are guided by the same statistical measures of success or failure as the MBA analysts, with some slight adjustments up or down for stronger or weaker beliefs. The rest of the movement in a company’s stock price we might attribute to exogenous events, whether macro (overall economy, regulation, etc.) ; micro (events affecting the general technology employed by the company, success or failure of competing programs); or company specific (failure or success of a program, management changes, acquisition rumors, etc. )

I think it is not that the DCF analyses are wrong, per se – they are actually insightful to a point but then employed in a way that would be useful only if constructing a portfolio of many investments such that the use of probabilities might lead to some meaningful expected outcome for the portfolio, with some reasonably expected band of upside and downside. The fact that these analyses, when employed to evaluate a discrete opportunity, sometimes lead to the avoidance of bad investments is no validation of the approach. They lead to the avoidance of M&A investments generally, and since drug development projects fail, any methodology that leads to a ‘no’ answer is simply betting with the house. Unfortunately, the avoidance of value loss is not the same thing as value creation. That takes guts and insight, not the input of information everyone knows into a model everyone uses.

Anonymous says:

Enjoyed the simplicity and clarity behind the thinking. That can be tough to do.

1) I am quoting one of my HBS professors who once said, “The greater the uncertainty, the greater the need for a financial model”. It is a view I ascribe to but certainly becomes more difficult for binary outcomes in drug development, especially in earlier stage assets. Scientific conviction is certainly key the earlier an asset is.

2) From my two years at HBS I have come to believe that financial numbers put discipline into intuition (Einstein – Intuition is a sacred gift and rationality a faithful servant). I hope companies are not making key decisions based on the actual NPV values. In fact as students we were made quite clear that those numbers need to be triaged with other factors such as what people are actually paying which becomes super-important in an auction-like setting (Medivation and 8 suitors) and multiples-based valuation.

3) An NPV also ring-fences (subject to assumptions) whether the price you are paying is too high – At some point the prudent thing to do is to walk-away but tough to do. We had an excellent case (from the aluminum industry) where the CEO was asking his head of corporate development to relax assumptions to make the model fit his view of the world.

4) I want to finish with this – In 2008, Lilly bought Imclone and BMS bought Medarex. The firms have taken two very different trajectories since. Would love to chat with Sigal about this. Did he actually seriously believe that Medarex had the key to the lock?

Anonymous says:

This has the potential to turn into a valuable teaching tool. Would love to see more examples of both good but especially bad. What are acquisitions that didn’t happen and did that violate your rules or did the non-acquirer make a mistake (eg vertex not buying pharmasset when it was more affordable and yet a bit less was known). How do we explain AbbVie’s two mega purchases? In both cases affordable (as you say for Celgene buying Receptos and Gilead buying pharmasset), and yet seemingly un-negotiated (Abbvie didn’t even call Acerta) and outlandishly high relative to all comps. Abbvie said that they were betting Ibrutinib would work in immunology, not just heme-onc, and yet there were no data at the time (and still little now). You say the lottery ticket is just as dumb for the rich person, so is that Abbvie? And I think the comment about the scientists needing to be sure deserves exploration. Without risk taking (i.e. Buying for the PoS, which presumably Gilead and Celgene did), won’t scientists clench up and set such a high bar that you’ll have a high positive predictive value but almost never do a deal (maybe even never); and when you do, it will be the oblivious deal for which 15 bidders will fight and you might almost regret winning (winners curse)? Yet if we permit portfolio theory thinking, we end up justifying spending on even early stage stuff… ie talking Pharma back into early stage R&D, which they have been bad at for a while. anyways… I know what it’s like to be inspired to share insights and then to ask for feedback. Hope this is helpful to you.

Anonymous says:

The issue is less about the NPV, DCF or IRR and more about the inputs to those models. The piece makes a great point about needing conviction from internal experts that are willing to underwrite and then drive the programs. If true P&L inputs, investment required, POS and risk of not doing (type 2 error) are assessed, challenged, incorporated into model then the output should be informative…[and] dispositive if the real work is done in an intellectually honest and robust way.

Taking account of the DS, DU and WIYD comes from analyzing upside potential, probability of success, capital at risk, and then a consideration of the losses or impact as a result of either not doing deal (and no on doing it) to not doing deal and someone else doing it. On the last it is the lost free cash flow one could have had versus doing nothing and the potential impact to existing base business if one does nothing. The latter could have regulatory questions associated with it

Anonymous says:

This piece should be taken further in one of two, or both directions:

• A Harvard case studies that looks at a number of successful and failed big bets and applies advanced financial analytics to flesh out your observations on RD, RU and WIYD
• A Harvard Business Review article on the key qualities of leadership in our industry.

On the latter, interestingly, while past experience influences courage (and, perhaps, foolishness), it is not fully determinative. Following a lost big bet, someone may be burned and shy away from taking another. Following a successful big bet, someone may become consumed with protecting the nest egg and refuse to take another.

Anonymous says:

I love it 🙂

Anonymous says:

This is great! I completely agree on the uselessness of NPV analyses and the importance of the “art” in making all M&A/in-licensing (and out-licensing) decisions. There’s just no mathematical approach that can replace the gut, wrinkles, and scars.

You should get Karen to let you publish this in Biocentury.

Anonymous says:

I really liked it, but thought the powerful prose and arguments were a bit let down by a rather wishy washy ending. Nothing personal: I was built up for a killer blow, but ended with nothing but a soft fart. The author really can write, if s/he ever fails, come and work in my shop. Ha!

Anonymous says:

Love this! Let me react over a glass of wine tonight or this weekend.

Anonymous says:

Total value created is the sum of de novo value created less value destroyed. The NPV/DCF way of thinking, when it fully takes over, is primarily concerned with protecting against value destruction…so much so that it gets in the way of de novo value creation resulting in less total value creation. Said another way, in this industry, while you certainly can spend yourself to death, you can’t save your way to glory.

Anonymous says:

Very interesting topic that doesn’t get enough explicit attention. Some of my thoughts below.

Re: Fully-Valued Late-Stage Assets – I agree : thereis no such thing as a “perfect” market. Assets are always valued in the micro (perceived management competence, company history esp re: hitting plan, overall pipeline/prospects, research analyst perspectives, etc) and the macro (competition, healthcare industry dynamics, US economy, etc) backdrop. As such, very reasonable people can disagree as to the value of said asset, given the different weights ascribed to these variables (influences not only POS of the asset but any premium/discount placed on it inherently because of these micro/macro factors). Moreover, an asset can be more valuable to a certain acquirer vs another….so one must consider value in the relative vs the absolute sense.

This dovetails into the Relative Upside / Relative Downside scenario analysis. I think that framing the discussion in terms of a binary outcome is useful – what do we gain if it’s a success vs. what do we lose if it’s an absolute flop. In terms of public markets investing, this is similar to a upside case vs downside case – weighing the relative odds of each and to help figure out whether there is asymmetric upside vs. downside. One can do a “scenario fan” analysis that takes into account a couple different decently likely scenarios (i.e. 25% chance of needing another Ph3 trail of $300M vs straight to registration); however, a binary approach can get you to the right directional answer. One must also recognize that blending RU and RD (or several other scenarios) results in an averaging case that is not realistic, and hence very flawed. Important not to lose sight of the forest when looking at the trees. Finally, your point about “what if you don’t” is important in that there is a broader strategic overlay that is hard to quantify but has real impact on whether to do a deal.but cannot easily be baked into profits/losses in an NPV.

Re: Courage of Your Best Scientists/Clinicians Conviction – Agree that people often default to 50/50 stance….in which case one should be very weary. A technical champion needs to make a firm stake in the ground and take professional/reputational risk (just as the business/finance team is taking risk). I’ve been in a situation where we were looking to acquire a late-stage drug into a defaulting (read: close to bankruptcy) spec pharma company. The decision to do the deal was based on our CSO-equivalent stating that there was an 80%+ chance of approval. We made the bet…and two years later sold the company for $3B.

I appreciate the thinking on the subject and agree with its general premise – valuing development-stage assets with inherent binary return profiles does not lend itself to any one “magical” numeric analysis. It’s is a combination of art and science!

Anonymous says:

I like this; it should be published somewhere. Two big thoughts, plus some comments within the document:

– Suggest losing the lottery section. Lotto part risks pulling reader into arguing with you on logic of whether a rich person vs. a poorer person should play the lotto. If you want to keep the analogy, you could pull into the RU/RD/WIYD section below – e.g. RU is same for both, but RD is very different. But think the analogy breaks down beyond that … e.g. wouldn’t WIYD favor a poor man playing the lotto?

– Suggest also tweaking to more of a “yes, and…” stance re: value of financial analyses. They’re not bad, they’re just woefully insufficient to get you to a decision. The kinds of frameworks and nuance you describe in here is what you need to get to a decision. But you still need DCFs and NPVs behind that, even in your framework. Can make the point without disparaging some of the key tools.

The piece caused me to think about more articles that could be written in teasing out the difference in risk taking between a rich man and a poor man, and on the WIYD part – too.

– If great things can only be accomplished by taking great risks, and putting large amounts of capital at stake, what kind of stomach does it take to take such risks? America’s voyage to the moon was financed by the government, and a lot of the really crazy science today is funded by the government in the name of the military, driven heavily by WIYD. In the private sector, RU ad RD rule, and we spread risks across portfolios to balance them.
– What are the circumstances that enable or impede large companies from taking big bets? Paradox of the RU being good, the RD being sustainable, but inability to act due to the slavery to quarterly earnings and unacceptability to investors of visible investments failing. I was thinking about comparison to riding a bicycle. On a bike, you peddle slowly uphill, and then fly downhill with much less effort. You don’t get to ride fast down the hill though unless you also put the effort in to climb up it; yet public companies are often punished if they break profile with a slower quarter or two, even if the promise of a nice downhill lies ahead. What companies have managed to get around this (e.g., maybe Gilead in your example) and what were the conditions that enabled them to do it.
– Big biopharmas manage portfolios of drugs, and VCs manage portfolios of companies. What are the benefits of doing the one vs. the other? A lot of this is psychological and cultural.

Anonymous says:

Value of MBAs

As the holder of a mere undergraduate degree, I can relate to skepticism of whether MBAs truly add value to biotech. Many of the smartest people I know have MBAs, but I think that is often due to self-selection rather than actual knowledge gained as part of the degree. I strongly disagree with the concept of business management as a technical specialty requiring a specific education. Leadership can be learned through many avenues and a broad set of backgrounds makes a management team less prone to blind spots. The main purpose of jargon is to quickly figure out who is in the club vs. not. I’m afraid being in a group of MBAs can sometimes be similar.

“Playing the Lottery”: Why Big Biotechs Should Bet More than Small Ones – But Don’t

While it may be less rational for a poor man to play the lottery, we know this is counter to actual behavior. Unfortunately, the poorest third of US households buy half of all lottery tickets. The top quintile by socio-economics status spend less per year on lottery tickets than the bottom (though participation of wealthy people tends to increase for those lotteries with enormous payouts). This may not be “rational”, but it happens. Similarly, the risk adverse behaviors that we see at some big companies with plenty of cash may not be rational, but we see it as well.

Why betting is not the same for a rich man as a poor man (or a big vs. small biotech):

– The intro reminded me of one of Bernoulli’s famous puzzle: “Somehow a very poor fellow obtains a lottery ticket that will yield with equal probability either nothing or twenty thousand ducats. Will this man evaluate his chance of winning at ten thousand ducats? Would he not be ill-advised to sell this lottery ticket for nine thousand ducats? To me it seems that the answer is in the negative. On the other hand, I am inclined to believe that a rich man would be ill-advised to refuse to buy the lottery ticket for nine thousand ducats.” Bernoulli proposed that the marginal utility of money is disproportional to wealth, thus, the poor fellow would be willing to accept with certainty a price much lower than the expected value of the lottery ticket. This can also explain why a small, cash poor, biotech may be willing to accept a buyout that slightly undervalues them with certainty vs. waiting to be worth either nothing or much more in future.

– Kahneman and Tversky’s Prospect Theory got at one of the challenges with Bernoulli’s theory: the “utility” of wealth depends on the reference point. For example: “Anthony’s current wealth is 1 million. Betty’s current wealth is 4 million. They must choose either a) the gamble: taking equal chances to end up owning 1 million or 4 million, or b) the sure thing: owning 2 million for sure. For both Anthony and Betty, option a would provide 2.5 million expected wealth while option b would provide expected wealth of 2 million, however, for Anthony this involves a gain relative to his initial amount, while it’s a loss for Betty. Because of this, Anthony is more likely to choose the sure thing while Betty is more likely to choose the gamble. When faced with bad options (as is for Betty), people are more likely to take their chances.” Perhaps it is the same for the small biotech that can either win big or lose big: you think you are worth at least $40M and Pfizer offers you $10M for 100% certainty, a $30M loss in your mind. You think you can wait a year and either have a 30% chance of being worth $0 or a 70% chance of being worth $100M. You wait!

– Kahneman and Tversky also did much to illustrate loss aversion: “we almost certainly dislike losing more than we like winning”. We see this risk-aversion in big companies quite often, much focus is placed on risk assessments and all the things that could go wrong. I think this risk aversion, combined with risk-loving behavior in the face of perceived losses, can explain some of the challenges we have seen with deal making in 2014-15: many small biotechs had perhaps had an inflated sense of value and were willing to wait for inflection points with potentially large payouts (especially in a hot IPO market) rather than take a perceived “loss”. On the buy-side, big biotechs were more nervous about the possibility of losing $5B (and face) than they are excited about the possibility of making $10B if successful.

The above can help describe why we see some of the behaviors we do but do not justify them as “rational” or, in the long run, good for shareholders. Kahneman argues that gains and losses should not be looked at on their own but rather in terms of states of wealth. If you asked a CEO if she would like a 50/50 bet to lose $5B or make $10B, she might say “no”. If she leads a big biotech with a $70B market cap and you ask if she would like a 50/50 bet to be worth $65B of $80B, she might reconsider (at least, these are the preferences expressed in Kahmenan’s experiments). If she leads a small biotech with $5B market cap and good independent growth prospects, she might be more inclined to say “no” to a bet that would leave her with a 50% chance of going bankrupt. This is consistent with the piece’s analytical approach, which suggests that a company looks at upside/downside in the context of what it means for the company, not just the gains and losses specific to the asset (I think this is most important for large deals).

Some Practical Thoughts
– At the end of the day, even the most complex NPV/DCF is all about judgment as well. We often associate quantification with objectivity and precision but that is absolutely not true. Every assumption that goes into an NPV/DCF (peak revenue, cost to launch, etc.) is only as good as the judgment of the person who helps make it.
– Importance of POS in assessing RD/RU
I agree that looking at RU and RD (as future states, not just for the asset) is a very valuable way to look at a deal. However, looking at RU and RD side-by-side without a POS can give an artificial sense of likelihood. People often default to “50/50” when they do not see an explicit probability associated with the options. Remember you will incur the costs to buy the asset and get to market at 100% but the return is less than certain – that POS assumptions is very important.
How do you avoid biases in assessing POS? Your scientists on the deal may not always have the most clear incentives. What if this deal would help elevate their stature and position? I suppose you could say that if you cannot trust your best scientists to make judgments for the good of the company they should not be working for you… But how do you know? Are there ways to account for these biases?
WIYD: to do this effectively, I think a management team needs to have a common view of the company trajectory and the challenges they are trying to solve. Doesn’t work if half of management thinks the sky is falling and the other half disagrees.
Your examples suggest that this is a good framework to identify good deals, but how about avoiding bad deals? Would Merck have backed out of Cubist if they used this framework? Would BMS not have done Inhibitex?
How does this work at a portfolio level, especially in the context of a strategy that involves multiple deals? This is something I have struggled with in biotech. There are many examples of industrial companies that have successfully acquired a platform co in a niche they wanted to enter and then tacked on smaller firms (Danaher, Honeywell, etc.). I think in biotech this is harder because you also have probabilities of success involved.

Anonymous says:

I have thought about these issues for a long time. One key consideration you need to introduce into this thinking is whether you are talking about a fundamental analytical approach to be deployed systematically, deal after deal, or an extreme one-off case that will never come up again.

Anonymous says:

I wrote multiple papers to my company’s leadership working on how NPV etc. deceive you. In multiple companies in my experience, not one (and I mean not one) projection of a successful or a failed compound EVER had a peak sales that was right – this is over hundreds of compounds. And yet decisions were made consistently using NPV, DCF, IRR, eNPV.

Anonymous says:

I enjoyed the piece and it came at a most opportune time for some of the internal discussions we are having. It manages to take a complex problem and boil it down to simple points, which is the best way, I believe people, to enable people to learn.

Anonymous says:

Excellent, thought-provoking read . . . and well worth expanding distribution through BioCentury, Nature, In Vivo, etc. Two basic comments:

1) Recognizing its well-documented limitations, NPV/DCF can provide useful information and context, all subject of course to the reliability of its various inputs (revenue and cash flow projections, discount rates, terminal/perpetual growth multiples, etc.). It is one of many accepted valuation tools, each of which has known limitations. Exclusive reliance on any one of them, whether quantitative or not, is simply a mistake. More importantly, none is a substitute for the cornerstone qualities of character, judgment, and guts, as is so poignantly exclaim in the last paragraph. We used a slide back in the ‘80s that illustrated the ridiculous range of potential value outcomes based on normal ranges for inputs into a standard NPV model.

2) The comment that the market perfectly values late-stage product candidates made me think about the standard efficient-market hypothesis, as taught in business schools, and whether nothing beyond its “weak” form applies to the pharma business. Might be worth thinking about that.

I also enjoyed the feedback you received over the weekend, particularly the comment that “Typically, there is much less risk, in the short term, to saying no vs saying yes. The most important learning for me is the reluctance most people have to making an affirmative decision where risk of being wrong is near term and measurable.” True that.

Anonymous says:

Excellent analysis!

I might add Vertex to your case analysis. VRTX knew that it needed a nucleoside inhibitor to add to its HCV franchise. Unwilling to pay for a high quality asset led to the demise of the whole franchise.

Anonymous says:

This is a great framework esp. as we begin to think about the decisions and opportunities that lie ahead for my company! There are always tools that help one understand and analyze the situation; however, you should never be a “slave” to the analysis and should not lose sight of the larger/other implications.

A couple of additional minor comments:

Your point re: the Rich man vs Poor man and the foolishness b/c of the “size of the bet” (use of high percentage of the assets) and not necessarily about whether a person was poor or rich (i.e. the rich man could also be foolish).

I like how you weave in the corporate strategy into the decision with your use of WIYD, but an aspect that is untouched is whether the opportunity is the driver or the strategy (i.e. in your Gliead example, should one view the Pharmassett opportunity as a “WIYD” or was the corp. strategy already set (retain anti-viral leadership), and the deal was a “no-brainer”.

Anonymous says:

Brightened my morning. Best concise summation of how to think about deals In the space, and the best part is that I can understand it. I especially like your approach of how to value and who to ask. I would suggest expanding the ‘so what happened’ part. You are intimately familiar with the outcomes but a student group (my current lens) is not. Could you expand on the consequences of each example? Lots of lives were saved by Gilead but other drugs were close. In both timing and profile. Why did they sweep? And it bought the team a high tide on which to exit. What was the post hoc “NPV”, e.g. (that might be fun.). And if it had been ordinary?

Love to hear about your take on Abbvie’s two deals for Pharmacyclics and Stemcentryx. What’s a truly bad deal?

The core is the notion of strategic value– no algorithm for that… Should there be? Love it.

Anonymous says:

Having just re-read this, I’m struck by how right it is. Perhaps our collective lesson is that to put this “art” into practice requires not just the judgement and conviction you cite, but a full management team that recognizes this simple wisdom. Or a subset of that team endowed with the accountability for such decisions. You can’t analyze your way to greatness just like you can’t cost cut your way to that. Hard lessons.

Anonymous says:

Agreed – if you have few (very different) potential outcomes, averaging makes no sense. You also can’t take the average of a left vs. right turn as you drive through town. You need to look at the individual outcomes like you said.

The DCF/ NPV approach would still make sense in the upside case. You want to make sure that the best case scenario gives you sufficient return for the risk you are taking – especially since the outcome is not certain.

WIYD is like the base case scenario of doing nothing – that’s what you compare the other two options against. For example, if the WIYD scenario looks pretty dismal, it might not take a fantastic upside to acquire the asset, because the RU (upside vs. WIYD) might be quite large, and perhaps the RD option is not that much worse vs. WIYD. This is a pretty important point that many people don’t take into consideration.

Actually I would think the 3rd option to assess is not WIYD but YNBA (your next best alternative). That was clear from your examples but I would make it more clear in the acronym itself. You need to compare RU and RD against YNBA.

Analyzing the RU and RD are pretty complex though because you won’t just be dealing with the acquisition cost but also any incremental development costs you have incurred and management distraction.

Anonymous says:

I think this points to the need for a much more holistic analysis of the business than just the transaction in isolation. I think the former occurs much more frequently than the latter.

Anonymous says:

First, agreeing that NPV/DCF’s are irrelevant to early stage assets, I do think they have their place for late-stage assets. Late stage assets can still be binary – either full loss or win, but what if the real question isn’t the POS of approval, but rather the likelihood of achieving certain commercial profiles? Here, having concepts of DCF/NPV and IRR can be objective tools for companies with larger portfolios to figure out their most efficient capital allocations. All that being said, it also can’t be the *only* factor, and here I agree with you that the relative upside/downside and the strategic role a transaction plays in a company’s situation is a primary consideration. And unfortunately, sometimes that upside/downside/strategic perspective is arrived at by a decision-maker from his/her own career perspective, and not from a company’s perspective! – could that be another explanation for why no other pharma was willing to pay the Pharmasset premium?

Anonymous says:

I love it.

Anonymous says:

The first page gave me a bit of an ice cream headache. And I understood the point about the metrics being insufficient for making business decisions. If they were sufficient, we wouldn’t need people to make decisions, we’d just have computers do it. Like many things, people try to make an exact science of something that still requires artistry and end up missing out on something. The other extreme is true, too: act only on emotion and you’re prone to devastating, obvious mistakes. The people who are most successful are those who can marry the two: have a solid scientific grounding and understanding of the field but also have the ability to act on their gut.

Anonymous says:

As it happens I’m in the midst of writing a piece about the “zone of ambiguity” in business decision making, and how misleading it is to think that analysis is enough to enable managers to make decisions “on the numbers”, so the piece is super useful.

Anonymous says:

It’s all about leadership.

Anonymous says:

Just saw this in BioCentury. Great piece. Reminds me of some terrific advice got from a mentor in the business. If you are thinking about buying an asset (or for that matter a share of stock) and all that differentiates your view on that asset is its valuation, then you are on a slippery slope. Valuations are not fixed, they fluctuate with time, risk, environmental and other factors, and can be susceptible to a subjective debate between any current or future owner or seller of the asset. If on the other hand you have a differentiated view on the fundamental attractiveness of the asset (the drug has a higher than appreciated probability of success, the drug will also work in a second indication, the population in need is far larger than appreciated) then you are on much firmer ground in making your decision. As you say, scientific insight is often what allows you to achieve that firmer ground.

Anonymous says:

What a great article. I loved all the philosophical comparators but of course the moral of the story of course is the best part. You really captured this perfectly. Thanks for sending a smile to my face with the fine print, that is very funny.

Anonymous says:

Loved it, Steve. Often the NPV is cast as helping to frame the debate. That is, it allows a transparent discussion about the model’s inputs and assumptions, including the PoS. Not untrue. But the analysis leaves out a bunch of critical considerations, as you rightly point out. True story: I saw an NPV kill a MED compound in the early 90s, pre-Viagara. Why? Because the market opportunity was based on the then-current treatments. After the math is done, it takes leadership and guts to make the call.

Erik Larsen says:

Kahneman and Tverksy’s “Prospect Theory” offers some nice concepts that can be framed in terms of common biotech investment scenarios, and may explain some typically observed behaviors.

I recently re-read the paper, and pulled out some of its ideas applied to biotech, in random sequence:

1. People tend to prefer higher probability of success (POS) opportunities over lower POS, even if they have the same NPV (and are even willing to incur a lower NPV). This would correspond to preference for investing in late-stage vs. early-stage assets (depends on the investor), as is a simple example of risk avoidance.

2. If all options have low POS, the opportunity with the largest potential pay-off is preferred, regardless of the actual POS. Makes sense – if we are unlikely to succeed, might as well shoot for the moon and see if you get lucky. People are expecting things to fail anyway.

3. There are a set of very interesting heuristics about how framing a decision differently can change people’s preferences, without any actual change in the probability of outcomes. I’ll just call out one, which is that in situations where an outcome depends on multiple uncertain sequential events (getting to IND, P1, P2, P3, approval), presenting the POS of various options as a composite will lead to a preference for the highest potential pay-off option (see point 2 above for low-POS options). However, if these options all have an equal and low POS of getting to P3, and that POS is pulled out separately in the presentation, the only difference that is seen is now getting from P3 to approval. That can reverse the preference, because now people will tend the pick the option with the highest POS from P3 to approval (point 1 above), which may be opposite to what has the highest potential payoff (point 2 above). This is really intriguing and could be a very valuable negotiation tool.

4. Value attributed to a payoff increases more slowly as the base value increases, and that the value of smaller sums is higher than the value of the summed $. Is it your experience that 3 projects worth $1B each are perceived as more valuable than a portfolio worth $3B?

5. Perception of probability is skewed – people overweight low-POS options (consider the lottery ticket). Even more interesting is that the weighting of an uncertain outcome and its alternative is perceived as < 1, which is nonsense mathematically. K & T interpret this as people being less sensitive to changes in probability than you would anticipate based on a strict application of expectation theory; e.g. increased the POS of an option from 50% to 60% doesn't fully increase the options' perceived value by 20%, as it "should". This is also interesting in consideration of how deals are valued and perceived, and what you should or shouldn't waste your time on in making a point about value.

6. As a final one (there's more in the paper), there is a difference in the risk people will accept when presented with an option to either accept a fair gamble vs. purchasing a gamble for a fair price. The first option would correspond to losing $100M at 90% POS or gaining $900M at 10% POS, while the second option corresponds to gaining $0 at 90% POS or gaining $1B at 10% POS at a cost of $100M. These are identical options, but the theory says that the latter option is typically less preferred, presumably because of the perceived fixed cost of $100M with no potential upside. The astute biotech investor (or seller) may use this his/her advantage.

Comments are closed.