Below Zero and Falling Fast: R&D Productivity as an Enterprise-Wide Crisis

Print Friendly, PDF & Email

Richard Evans / Scott Hinds / Ryan Baum

203.901.1631 /.1632 / .1627

richard@ / hinds@ /

twitter.jpg @SecSovHealth

October 11, 2011

Below Zero and Falling Fast: R&D Productivity as an Enterprise-Wide Crisis

  • We estimate current returns on R&D spending of -7% – before costs of capital – for a typical R&D portfolio (typical in terms of # of projects, mix of projects by phase of development, and mix of projects by large v. small molecule)
  • Critically, we find that the industry’s rapid pace of real pricing growth – in particular for small molecule products sold in the US – accounts for half of total returns on R&D spending
  • Because we anticipate an erosion of US real pricing power, we now also anticipate that returns to R&D spending will deteriorate further, in all likelihood at an accelerating pace
  • Managements have viable options for rescuing R&D returns, but these are not being adequately pursued for two apparent reasons: the problem (poor R&D productivity) is defined too narrowly, and targeted productivity gains are far too modest
  • R&D productivity is an enterprise-wide measure – the COGS and SG&A costs of commercializing a discovery are just as impactful to R&D returns as the time and costs associated with bringing the product to approval. Framing poor R&D productivity as a problem that exists narrowly within the borders of R&D operations appears common, and severely limits options
  • Recent cost-focused restructurings are benchmarked to income-statement measures (e.g. margins, EPS), but current-period (or even 5+ year forecast period) income statements cannot ‘see’, and accordingly cannot reflect the scope and scale of the R&D productivity problem. Current-period profits are the result of R&D long since spent, and the bulk of returns to current-period R&D spending are perhaps a decade in the future
  • By comparing the present value of future profits for the typical project to the cumulative development costs (projected to the point of approval) for the typical project – as we do in this research note – the firm can produce a meaningful estimate of current R&D productivity
  • We conclude that R&D productivity is deteriorating at a potentially accelerating pace, and that firms are capable of rescuing their R&D productivity, but we can think of none who are yet on track to do so
  • We continue to recommend therapeutics portfolios built solely around companies with pending or recent product approvals (i.e. positive idiosyncratic risks), in large part for the reason of avoiding deteriorating systematic (e.g. R&D productivity) risks


Research and development is the starting point of value creation in the biopharmaceuticals industry. Discovering, developing, and commercializing innovative products requires tremendous capital, and returns from R&D must exceed the cost of this capital if the industry is to remain viable over the longer-term

Costs of capital are well established, but estimates of R&D returns vary widely, as do the means of producing these estimates. In the past, we’ve relied on a simple comparison of Yr0 R&D spending to Yr10 adjusted[1] net income. This method is advantaged by the fact that it is simple and comprehensive, but disadvantaged in that it can only speak to directional trends (as opposed to specific estimates of R&D productivity for specific periods); it cannot speak to why R&D productivity is changing; and, its most recent estimate is always ten years old. Despite these limitations, work conducted under this method (limited in scope to large cap pharmaceuticals) paints a convincing picture that returns on R&D are falling, and that the spread between R&D returns and cost of capital is by now – assuming the trend has continued – negative (Exhibit 1)

We’ve developed an improved model of R&D productivity that offers three advantages: 1) the model is both simple and comprehensive; 2) by using contemporary inputs, the estimate of R&D productivity is more up-to-date; and, 3) through scenario testing, the model speaks to where R&D productivity may be heading, and also to what alternatives managements may have for improving the trend

The Model in Brief

In simplest terms R&D productivity is a newly approved product’s net economic value (to the producer) divided by the cost of developing that product to the point of approval. The formula and its inputs are straightforward, however finding accurate values for these inputs is challenging

Before approval, products progress through various phases of discovery and development. Each stage has associated cash costs, risks, and ‘time costs’. Because peer-reviewed and empirically-grounded estimates of costs, risks, and time in phase are readily available, we can model the average cost of ‘creating’ an approved product as the cumulative cost of completing the entire discovery, development, and approval process

Our cost calculations start at the end of preclinical testing, just before a candidate compound enters human trials. The cost we assign to completing research and preclinical testing is the specific cost of getting that specific compound to that specific point, plus that specific compound’s ‘share’ of costs for compounds that failed in preclinical testing. For example, if we assume 1,000 compounds fail at some point before the end of preclinical for every compound that completes preclinical testing, then – at least for the purposes of our model – the amortized cost of completing preclinical testing includes the cost of the 1,000 failures. From this point forward, risks are dealt with in terms of the odds of progressing to each subsequent phase; this risk measure is combined with the (average) cost of each phase and the (average) time spent in each phase to produce an estimated cost for successfully progressing to the subsequent phase. These values are quite different for large and small molecules, so we model large and small molecule development costs separately. This method produces cost estimates by phase and type of molecule (i.e. large v. small); and critically, this detail allows us to estimate R&D productivity at the portfolio level, by simply building model portfolios with representative distributions of products across molecule types and developmental phases. Exhibit 2 is a figure summarizing costs, time, and risks (expressed in terms of the probability of either progressing to the next phase or of completing the entire development and approval process) across the development timeline for both large and small molecules

For simplicity, we assume that firms immediately sell their products the moment regulatory approval is received; and, we assume the payment received is exactly equal to the present value of an average product’s future profits[2]. To estimate the value of future profits, we build simple net income models for average large and small molecule products. Using actual historic US and ROW sales values for 1,513 products across the 11 most recently completed calendar years, we build nomograms of sales by year since launch, by molecule size. We break out the observed contribution of various inputs (e.g. real pricing, volume) to these nomograms, so that we can both measure the effect of these inputs on historic R&D returns, and model the effect of any future changes to these inputs on future returns. We benchmark our assumed net margins to the net margins we observe at the whole-firm level for predominantly large or small molecule manufacturers, and adjust net margins for tax-effected R&D spending (which is added back to net income[3]), and also for stage of product lifecycle (we assume lower margins early, and higher margins later). Exhibits 3 and 4 are line graphs of the percent of peak sales achieved by years since approval and assumed net margin for large (Exhibit 3) and small (Exhibit 4) molecules. Exhibit 5 is a table summarizing key product forecast variables (peak sales in 2011 dollars, % of sales by geography, contribution of real pricing by geography, average net margin, and average PV of future income) for both large and small molecules

Because firms invest in portfolios of R&D projects, we take two further steps to organize costs and returns across representative portfolios. Specifically we assume the portfolio is ‘typical’ in terms of: 1) the total number of projects; 2) the distribution of projects across phases and molecule types (i.e. large v. small); and 3) the mix of projects discovered internally, and projects that were in-licensed[4] (Exhibit 6)

We discount our projections using estimates of a typical firm’s R&D-specific costs of capital. Industry WACC levels are quite low, however for many firms costs of capital are heavily influenced by the existence of predictable cash flows that are multiples of debt service levels. Accordingly firm-wide WACC figures often meaningfully under-estimate the risks, and thus true capital costs, of the R&D process. We rely on empirically-anchored and rationally-derived costs of R&D capital from peer-reviewed literature[5]

A Current Estimate of R&D Productivity

For a typical portfolio we estimate a return on R&D spending of -7% (Exhibit 7). We caution against over-interpreting the specific percentage outputs of the model; despite the care used in its construction and in sourcing inputs, we conservatively assume the model has an error rate of +/- ‘several’ percentage points. Nevertheless having produced a return estimate of -7%, before considering costs of capital, we’re comfortable concluding that returns on R&D spending for a typical portfolio are at best very small, and probably negative. Layering in the matter of capital costs, we’re even more comfortable concluding that R&D returns are below capital costs for the typical portfolio

Returns are much better for large molecules (we estimate +16% across a large molecule only portfolio) than for small molecules (we estimate – 22% for a small molecule only portfolio); and, returns also are much better for late-phase projects than for early-phase (and preclinical) projects (Exhibit 7, again; and Exhibits 8 and 9). Large molecule returns appear to be superior for several reasons, the most important being higher average peak sales values, larger average net margins, lower developmental risks, and greater ROW pricing power (Exhibit 10). Early projects have lower returns than later projects, simply because the cumulative effects of time and risk (which reduce the present value of a possible future approval) far outweigh the tendency of per-phase costs to be (somewhat modestly) lower at earlier stages of development

The superiority of returns in large molecule discovery and development does much to explain the industry’s growing infatuation with large molecule R&D. However, the associated fact that capital committed to large molecule R&D may be expanding faster than the availability of attractive large molecule concepts signals these returns may fall. Separately, we clearly recognize the complex irony of negative returns in early development and positive returns in late development. Interpreted simply, this suggests that early development is value-destroying and should not be done; however, profitable late-stage projects cannot exist unless they were once (in expected value terms) unprofitable early stage projects. At least part of the answer to this chicken-and-egg conundrum is a well-balanced R&D portfolio, whose late stage positive values outweigh its early-stage negative values. Unfortunately at the moment, the typical portfolio is not well-balanced, and returns are almost certainly below costs of capital. This begs the question of whether returns can be brought back into attractive territory; before attempting this question it’s important to first understand where these (presently negative) R&D returns appear to be headed – i.e. are things getting better, or are things getting worse?

Which Direction are R&D Returns Headed?

Exhibit 11 shows the effect that a single percentage point change in input assumption(s) has on our estimate of R&D returns at the ‘typical’ portfolio level. Most importantly, we see that R&D returns are more sensitive to the real rate of US small molecule price change than to any other input. Other than the dominant effect of US pricing on small molecule returns, large and small molecule portfolios have similar sensitivities to inputs, with estimated R&D returns being most reactive to (in descending order): total development costs, total time in development, and total development risks. Total development costs (costs/NME) have been growing exponentially; and, time in development has been reasonably static for small molecules, but growing for large molecules, thus total average time in development has been on an upward trend. Risks also have increased: for example the odds of a phase I compound reaching market fell significantly between 1996 – 2001 and 2002 – 2007 (Exhibit 12). While it’s quite clear that the average product of costs x time x risk has recently been on an upward trend, it’s less clear precisely what is behind these trends and whether these trends will continue, and without such an understanding we’re unwilling to simply force an assumption of rising costs, time, and risks into the model. Accordingly the model conservatively assumes a leveling off of the cost, time, and risk trends; if this assumption is incorrect then actual returns would be worse than our estimates

We’re far more convinced that we have a handle on middle- to longer-term prospects for the industry’s pricing power; specifically we believe that US real pricing power eventually is lost. It’s incredibly clear that this has a very large – almost defining – effect on R&D returns. Most simply, we might rely on the obvious argument that no industry can sustain positive real pricing in perpetuity. More usefully, we might point to mechanisms such as the Independent Payment Advisory Board (IPAB) on the government side (see
), and a shift from co-pays toward co-insurance on the commercial side, as means by which real pricing power might be lost. Even closer in, we see very high odds of the Budget Control Act’s so-called ‘super committee’ raising effective drug (and other innovators’) rebates (see
); and also expect real drug pricing to decelerate to or toward CPI – at least for a brief period – with the opening of a general election year in 2012 (see

Against the backdrop of a ‘typical’ portfolio (in terms of the mix of large and small molecules across phases of development), we see the enormous effect of US real pricing power. We estimate historic US real pricing of 4.2% for a typical portfolio across our look-back period of 2000 – 2010; and, at this assumed real pricing level we measure a -7% return for a typical portfolio. Returns are highly sensitive to pricing power; a very slight increase in the real pricing assumption to 4.5% brings the estimated R&D returns on a typical portfolio to 0%, i.e. ‘breakeven’ (ignoring costs of capital). Unfortunately real pricing power is more likely to fall than rise, and the impact on R&D productivity is quite simply dramatic. All else held equal (e.g. costs, time, risks by phase, mix of large and small molecules, mix of in-house and in-licensed, and distribution of projects across phases), loss of US real pricing power reduces estimated R&D returns by more than half (Exhibit 13). Because we’re convinced US real pricing power is fading, and because US real pricing power has a dominant effect on R&D productivity, we’re convinced that R&D productivity is on a (potentially accelerating) downtrend

What Can Managements Do?

Take an Enterprise-Wide Approach to Defining and Addressing R&D Productivity

Return on R&D spending, or interchangeably, “R&D productivity”, is an enterprise-wide measure, i.e. the manufacturing and sales costs of commercializing an approved product are just as relevant to R&D returns as the costs, time, and risks involved in development. Accordingly, managements’ options for improving R&D productivity also are enterprise-wide; focusing narrowly on the discovery and development process is, we sense, a common mistake. Pursuing ‘non-R&D’ efficiencies for the purpose of raising R&D returns also is essential because managements may be more able to affect productivity outside of the discovery and development process than inside. The logic is simple: inside the discovery and development process, variables that determine R&D productivity (costs, time, risks) are heavily (though not entirely) influenced by systematic forces beyond managements’ control. Managements are price-takers in many of the critical discovery and development transactions (e.g. compound in-licensing, the costs of attracting world-class talent in narrow disciplines, and/or the costs of patient / investigator access in crowded clinical trial spaces); time in development is heavily affected by regulatory considerations that are beyond managements’ control; and, high absolute risk levels are a natural feature of the life sciences. In contrast, we believe managements have comparatively high levels of control over manufacturing, marketing, and administrative processes, and accordingly must make efficiency gains in these processes a core feature of efforts to improve R&D returns

The need to find efficiencies on an enterprise-wide basis is compounded by the sheer scale of the R&D productivity challenge. Bear in mind that our estimate of -7% R&D productivity assumes that compounds in development today have sales trajectories that include historic levels of real pricing power. More realistically, we believe real pricing power will be lost in the middle- to longer-term, which has profound effects on forward-looking R&D returns. If we assume zero real US pricing power for all products currently in development, our estimate of R&D productivity (i.e. the expected value of future profits from current R&D spending) falls from -7% to -56%, before considering costs of capital. To further highlight why R&D productivity efforts must span the entire enterprise, consider just how much managements would have to alter R&D operations to offset the loss of real pricing power. Of the three main determinants of productivity within R&D operations – costs, time, and risks – we believe managements can most readily affect costs and time, since discovery / development risks are (even more) heavily influenced by factors beyond management control. A hypothetical management team committed to offsetting the R&D productivity headwinds from real pricing power (to say nothing of making R&D returns positive), and focusing only on time and costs in R&D operations, would have to reduce both average time in development and average cost per phase by 36% – a wholly unrealistic goal. We are utterly convinced that managements cannot defend current levels of R&D productivity against the loss of real pricing power – much less improve R&D productivity – by narrowly focusing on time and cost efficiencies in the R&D process. R&D productivity efforts must be enterprise-wide to have any realistic chance of bringing R&D returns above costs of capital

Get an Accurate Measure of R&D Productivity – Which Cannot be Found in the Income Statement

After recognizing R&D productivity as an enterprise-wide challenge, we believe the most important next step is for managements to more accurately / realistically measure the magnitude of their organizations’ R&D productivity headwinds. The industry’s most recent era of cost-focused restructuring has been measured against the income statement, which is a useless reference point in light of the multi-year time periods across which R&D productivity unfolds. Current-period revenues and (especially) earnings are predominantly the result of R&D spending that appeared on income statements many years before. And, the lion’s share of economic value produced by today’s R&D spending will not meaningfully affect earnings until quite some years in the future. Despite this, current-period cost structures – including of course current-period R&D spending – generally have been scaled to allow for an acceptable level of current-period net income. By scaling expenses to current-period (or even strategic planning period) earnings targets, the bar for organizational efficiency / productivity is set far too low; accordingly any restructuring effort that hopes to make returns on R&D spending positive cannot use the income statement as its productivity benchmark. Instead, R&D productivity should be measured and managed to by comparing the probable present value of future product approvals to the present-day cost of discovering and developing those products – as we have done here

Think Like a Commodities Producer – No Feasible Productivity Gain is Too Small or Too Difficult to Pursue

Considering a typical biopharmaceuticals firm against the backdrop of realistic R&D productivity measures, we see multiple opportunities for change. And, given the scale of the R&D productivity challenge, we believe that practically all feasible opportunities to gain efficiency will have to be addressed. The following opportunities are those we feel are most often either ignored, or pursued with an insufficient level of focus or intensity

R&D time and costs

Reducing average time and costs by R&D project and phase of development is essential, and according to our sensitivity analysis is a logical priority. However as we’ve emphasized, R&D time and costs are heavily affected by systemic risks beyond managements’ control, reducing the potential savings; and, it’s unrealistic to expect that R&D productivity can be made positive by addressing time and costs alone

We believe an essential area of focus is in early discovery, where costs are substantial, but the value of associated returns is low because of the time, costs, and risks that stand between this phase of the process and the profits from an approved product. Our strong belief is that firms cannot afford to retain large staffs and bear associated costs in developmental disciplines where they are not world-class – but that many firms do

Average development times should be attacked both in terms of time to success and in terms of time to failure. Traditionally we think of time in terms of how long it takes a product to move from a given stage of development to approval; however, given the high attrition rates inherent in drug development, reducing the time failed products remain in the portfolio is a sizable source of potential efficiency. Exhibit 14 shows the considerable impact of reducing the average time a failed project spends in the development portfolio

Portfolio-level R&D risks

We’ve argued that developmental risks may be largely inherent and thus difficult to lower on a project by project basis; however managements can still affect the cumulative risks in development portfolios by: 1) using straightforward portfolio models to calculate risk at the portfolio level; and 2) using this information to balance the portfolio’s risk / return characteristics[6]. Even basic portfolio-based planning processes offer the added benefit of improving the organization’s ability to forecast the longer-term economic value of its development portfolio – an essential step to measuring R&D productivity. Our experience is that even leading biopharmaceuticals firms fail to use off-the-shelf tools to measure, much less manage, portfolio-level risks

Smaller, more flexible (enterprise-wide) cost structures

The pending loss of real pricing power not only reduces R&D productivity – thus compelling a smaller cost base; reduced real pricing power also leads to revenue volatility[7], which argues in favor of a substantially more flexible cost structure. We acknowledge that ‘smaller and more flexible’ has become a re-sizing catch phrase in some of the industry’s larger companies; however we sense the scale of how much smaller and how much more flexible has been benchmarked more to the income statement than to an accurate forecast of what costs can be carried in light of probable future product values – i.e. we believe these firms are not becoming small and flexible enough

Broadly, we believe most firms can reduce the cost base and gain flexibility more than has been done by taking such steps as:

  • exiting discovery disciplines in which the firm is not world-class (in terms of innovation and thought leadership);
  • narrowing the development footprint to functions that are strategically essential (e.g. project management, regulatory affairs) by outsourcing functions where the firm’s cost-adjusted operating effectiveness is less than world-class, or where high utilization cannot be maintained;
  • outsourcing small molecule manufacturing[8]; and,
  • fundamentally re-thinking the essential scale of SG&A functions, and re-defining basic SG&A processes to fit within this essential scale

Look for and capitalize on mis-valuations at the project and portfolio levels

Obviously the value of an average early-phase project is less than that of an average late-phase project, where most costs and risks have been carried, and the time to payoff is short. However we find that early-phase projects, at least for the moment, on average have negative values. We also find at least some evidence that deal costs for in-licensing developmental projects may in many cases be less than the true economic cost to the innovator of having developed the project to the point at which it was licensed. We’re not entirely convinced this is correct on average, but we’re sufficiently convinced it’s true on enough occasions to warrant firms adopting a far greater, and far more opportunistic, in-licensing focus

Similarly, we’ve shown elsewhere that firms’ share price values are a combination of tangible (anticipated future earnings from on-market or near-to-market products) and intangible (anticipated future earnings from pipeline products whose commercial features are not yet in plain view) values. And, because product pipelines are a large percentage of total enterprise value, and also because the capital markets are relatively inefficient at valuing intangible assets, opportunities continually arise for firms – and their development portfolios – to be purchased at net costs well below the cost of creating a similar portfolio

Investment Relevance

Our belief that R&D returns are below costs of capital has for some time been the main foundation of our bearish thesis on large cap pharmaceuticals, and this thesis is confirmed by our updated model of R&D productivity. What’s more, the model gives us a new appreciation for the link between US real pricing power and productivity; and, since we believe US real pricing power is fading, we now believe that R&D productivity for ‘traditional’ small molecule portfolios (still the cornerstone of most large cap pharma discovery and development) deteriorates further, and that this deterioration may unfold at an accelerating pace. Unless efforts to gain productivity grow at a similar pace, we believe that even the modest earnings expectations reflected in current large cap share prices cannot be met

In contrast, we’ve been considerably less bearish on biotechnology companies, primarily because the impact of generic competition here is much smaller than is the case with large cap pharmaceuticals – a distinction we expect will remain important even after follow-on biologics reach their full potential. Comparing the R&D productivity of large and small molecule portfolios, we find that large molecule portfolios offer superior (and positive, and above cost of capital) returns as compared to small molecule portfolios; and, that large molecule portfolios appear far less susceptible to the effects of eroding real pricing power[9]. Combined, these findings add significantly to our (at least relative to large cap pharma) pro-biotechnology bias

Despite our conviction that biotechnology is far better positioned than traditional large cap pharmaceuticals, this distinction alone is an insufficient basis upon which to build a portfolio of attractive therapeutics stocks. Instead, we favor portfolios of therapeutics stocks with approaching or recent regulatory action dates on major products. The immediate gain in earnings power associated with a major product approval offers the best opportunity to offset the general secular trend of eroding earnings power among therapeutics companies broadly; and, the somewhat predictable patterns of share price performance around these regulatory action dates offer an opportunity to both capture the share price gains associated with new product potential, and mitigate exposure to the share price risks associated with non-approval decisions. For details, please see: “A Simple Formula for Drug (and Biotech and Spec-Pharma) Stock Selection” Sector & Sovereign Research LLC, Sept. 9, 2011

  1. We credit tax-effected R&D spending back to net income, and debit from net income the amount of profits that would have been expected if the same volume of product had been sold without patent protection
  2. We see the shortcoming in this simplifying assumption – specifically, by assuming the R&D firm is paid 100% of the future value of product-specific income, we offer no returns to the commercial operation that manufactures and sells the product, and in so doing effectively assume that the capital needed to make and sell the product is ‘free’
  3. In other words the profits attributed to any product assume that the firm manufacturing and selling the product conducts no R&D. We of course recognize this is rarely if ever the case; however this construct does give a more accurate representation of the product’s economic value – manufacturing, sales, and administration are necessary expenses associated with commercializing a product; R&D is not
  4. There are two major effects of in-licensing on R&D returns, and both are challenging. First, there is considerable controversy over whether in-licensed projects currently have higher or lower probabilities of success. Having no ready means of settling that controversy for ourselves, we assume no difference between in-licensed projects and projects initiated in-house. Second, there is considerable controversy over whether in-licensed projects offer higher or lower returns than internal discoveries. Plainly in-licensed projects offer lower future profits because of up-front payments, milestones, and royalties; however some evidence suggests that the cumulative value of these payments is, on average, less than the cost of having developed the same compound to the same point internally. Our strong sense is that the market for intellectual property favors the seller. For our model, we assume manufacturers: 1) in-license compounds for an up-front cost equivalent to developmental costs incurred up to the point of licensing; 2) pay all future development costs; and, 3) pay royalties equal to one-quarter of expected future net profits attained through year of peak sales
  5. Scott Harrington, “Cost of Capital for Pharmaceutical, Biotechnology, and Medical Device Firms” 13 Nov 2009
  6. E.g. by changing the blend of risks across projects in the portfolio, or by re-weighting risk exposures through partnerships (for example owning 50% of 100 projects instead of 100% of 50 projects)
  7. For most of the industry’s modern history – the exceptions being periods of high political risks – real pricing power has been both a large and a discretionary component of overall revenue growth. More simply, real pricing has served to both raise and ‘smooth’ the industry’s rate of revenue growth. As real pricing power fades, the industry’s revenue growth should become both slower and more volatile
  8. Firms typically need several distinct types of small molecule manufacturing facilities, making high asset utilization across all of these facilities very difficult to achieve. And, because in-house manufacturing is shielded from competitive pressure, we believe vendor-owned small molecule manufacturers – forced to compete – would evolve overall levels of cost and quality more rapidly. For a variety of technical reasons large molecule manufacturing is an entirely different story where we do not support outsourcing
  9. The general pattern has been for small molecules to enter the market priced at or near the level of some prevailing benchmark, then for price to inflate subsequently in real terms for most of the product’s lifecycle. In contrast, large molecules tend to enter the market at very high real prices, with very little real list price inflation until much later in the product’s lifecycle – if at all. Accordingly any loss of the industry’s freedom to change the price of its products periodically in real terms – which is exactly the type of change we anticipate – falls more heavily on small molecules than on large molecules. We recognize that large molecule launch prices have been growing dramatically, and acknowledge that these prices cannot continue growing in real terms indefinitely – however our model only assumes that large molecule launch prices grow between now and the date of launch of products in development, at the rate of general inflation (i.e. CPI). We also acknowledge that various threats exist to large molecule pricing, including in particular the Independent Payment Advisory Board; however we believe that the scale of large molecule pricing threats is considerably smaller than the scale of real pricing threats facing small molecules
Print Friendly, PDF & Email