PFE decides to shrink; HHS ends AWP; and, drug pricing hits a speedtrap

Print Friendly, PDF & Email

Richard Evans / Scott Hinds

203.901.1631 /.1632

richard@ /

February 11, 2011

PFE decides to shrink; HHS ends AWP; and, drug pricing hits a speedtrap

  • PFE’s R&D / sales will fall to 10.5% in 2012, down from 13.9% last year. Strictly speaking, with R&D returns < WACC, the only ‘right’ R&D / sales ratio is zero. Assuming R&D returns can be >= WACC (we believe they can), companies need R&D / sales ratios of 15% simply to hold the real value of sales constant. All else equal, and assuming R&D equal to WACC, PFE’s reduction in its R&D / sales ratio shrinks the present value of the company’s long-term earnings by roughly one-third
  • On February 3rd, HHS Secretary Sebelius announced plans to publish a survey of drug acquisition costs by the end of this year. This obviates the need for AWP, which is the benchmark used by commercial contracts; presumably these contracts now must be re-written. HHS-provided acquisition cost data would give plan sponsors the details on pharmacies’ generic acquisition costs that they’ve been lacking; we expect 2012 generic dispensing margins to fall as a result. PBMs (MHS, CVS, ESRX) are most negatively affected, followed by drug retailers (WAG, RAD), then drug wholesalers (CAH, ABC, MCK)
  • List price increases for US drugs tend to decelerate in presidential election years, especially when prices are growing rapidly in the year pre-election, as is presently the case. This is particularly important in light of the fact that real pricing is now a more dominant component of revenue growth than it’s ever been, making its potential loss all the more impactful. Manufacturers with weak volume / mix trends are most at risk, e.g. GSK, PFE, JNJ, SNY, and MRK. Wholesalers (CAH, ABC, MCK) also are at risk as the buying margin benefits of large real price gains may be lost abruptly
  • We provide an updated drug stock selection screen, adding specialty pharma and biotech to our existing screen of large caps. We select for low sales-weighted average product age, high percentage of sales attributable to biologics, and high ratios of phase III or filed products to current sales. Predictably large caps fare poorly, which argues in favor of more actively managed portfolios of small – mid cap names

R&D productivity – does cutting R&D as a % of sales make sense, and how much R&D spending is required simply to maintain current sales?

PFE recently lowered its guidance for 2012 R&D spending to $6.5B to $7.0B on sales of $63B – $65.5B; this drops PFE’s R&D to sales ratio to 10.5%, down from 13.9% in 2010

Whether others follow is unclear, though PFE’s actions are sufficient to raise the questions of whether cutting R&D spending makes sense, and of whether the industry is conducting research at a rate sufficient to maintain its present size

In aggregate, the large cap drug manufacturers’ R&D productivity appears to be below cost of capital (Exhibit 1). We estimate R&D productivity by comparing YR10 net income to YR0 R&D spend; an obvious limit of this method is that we can’t measure past 2001 R&D spend. Nevertheless the trend is clear, and being convinced that R&D productivity has almost certainly continued to fall since 2001, we’re likewise convinced that R&D productivity is now below WACC. Spending on projects that return less than WACC simply prolongs the firm at the expense of shareholders, which suggests an obvious answer to the question of how much should be spent on R&D: strictly speaking, firms should spend nothing. Thus spending 10.5% of sales on projects with returns below WACC isn’t a positive, it’s simply less of a negative than spending 13.9%

Thus unless we believe R&D returns are above WACC, the question of how much should be spent is mooted. We believe R&D can return more than cost of capital, though the necessary changes are extensive. Rather than belabor them here, we’ll assume for the moment that the companies can get R&D returns back above WACC, which at least enables the question of how much R&D is necessary

The simplest way to approach this is to think in terms of sales returns on R&D, i.e. YR10 sales as a function of YR0 R&D spending. Since 1983, sales returns on R&D spending have been falling (Exhibit 2), consistent with the falling R&D productivity we observed when we looked at things on a net income basis. Importantly, real pricing power in the United States was a major contributor to

historic (global) sales returns on R&D spending, but we do not believe that real pricing can play such a role in the future. Unadjusted for the contribution of US real pricing gains, the actual sales return on historic R&D spending fell from roughly 41% in 1983 to 28% in 2000. If we eliminate the contribution of US real pricing gains to these global returns, then the figures fall to roughly 37% in 1983, and 24% in 2000

We hold R&D productivity constant at its last-observed value (2001), and solve for the R&D investment required to grow the industry’s top-line at roughly the long-term rate of inflation (2.5%), under either of two basic assumptions: US drug list price growth continues at its historic real pace; or, real price growth falls to zero (i.e. drug list prices grow at CPI). Under the former assumption, the industry could hold the real value of its sales constant by spending roughly 11% of sales on R&D, well below the current average of 14%. Under the more realistic assumption that drug pricing in the future largely mirrors CPI, we estimate the industry would need to spend roughly 15% of sales on R&D, which is slightly above current (and forecast) spending

Thus in rough terms, if we assume 2001 R&D productivity levels are sustained, current R&D spending is nearly sufficient to maintain the real value of industry sales. More realistically we believe R&D productivity has continued to fall, which argues that current industry levels of R&D spending are insufficient to maintain current sales in real terms, i.e. the industry is spending at least a little bit below its replacement rate

Pfizer, at 10.5%, is unambiguously below the replacement rate of R&D spending; all else equal, Pfizer’s decision to drop the R&D ratio from 13.9% to 10.5% is an explicit choice to reduce the company’s size by roughly one third

Most interesting of all is whether PFE’s stock should be worth more or less as a result of the adjusted R&D spend, i.e. whether the near term net margin gains can offset the effect of reduced sales and earnings longer-term. Again assuming that R&D returns can be made at least equal to WACC, it turns out that raising PFE’s R&D spending to the replacement rate (15%) is a far better decision (in terms of shareholder value) than cutting R&D to 10.5%. At a 15% R&D / sales ratio we would expect 2.5% annual revenue growth, though net margins fall slightly as a result of the increase in R&D spending versus the 2010 base (13.9%). At a 10.5% R&D / sales ratio we would expect annual sales erosion of around 75bp, though net margins expand as a consequence of reduced spending. All else held equal, the present value of earnings from the 15% R&D spending scenario is 1.37 times greater than the present value of earnings in the 10.5% R&D spending scenario, and 1.49 times greater than a scenario in which R&D is cut altogether[1]

HHS signals a shift of the Medicaid drug pricing benchmark from AWP to actual acquisition cost this year; generic dispensing margins are likely to compress as a result

Our bearish view of the drug trades generally, and PBMs particularly, has multiple components; perhaps the most critical being our expectation that generic dispensing margins ultimately will compress, and that this will come about as a consequence of plan sponsors having greater visibility into pharmacies’ generic acquisition costs

On February 3rd, in a letter to governors addressing concerns over Medicaid costs, the HHS Secretary outlined plans to publish a survey of actual drug acquisition costs by the end of this year:

Excerpt of February 3, 2011 letter from HHS Secretary Sebelius to state governors (emphasis added):

“… We are committed to working with States to ensure they have accurate information about drug costs in order to make prudent purchasing decisions.  As recommended by States, the Department is undertaking a first-ever national survey to create a database of actual acquisition costs that States may use as a basis for determining State-specific rates, with results available later this year …”

And from the letter’s attachments:

“… there is still strong evidence that many State Medicaid agencies are paying too high a price for drugs in the Medicaid program.  Recent court settlements have disclosed that the information most States rely upon to establish payment rates is seriously flawed.  As a result, the major drug pricing compendium used by Medicaid State agencies will cease publication before the end of 2011, and States must find a new basis for drug pricing.  We will work with States to help them manage their pharmacy costs and ensure their pharmacy pricing is fair and efficient:

    • Provide States with a new, more accurate benchmark to base payments.  A workgroup of State Medicaid directors and State Medicaid pharmacy directors has recommended a new approach to establishing a benchmark for rates, namely, use of actual average acquisition costs.  Alabama, the first State to adopt use of actual acquisition costs as the benchmark for drug reimbursement rates, expects to save six percent ($30 million) of its pharmacy cost in the first year of implementation.  However, it is difficult and costly for each State to create its own data source for actual acquisition costs.  States have recommended a national benchmark.  In response, CMS is about to undertake a national survey of pharmacies to create a database of actual acquisition costs that States may use as a basis for determining State-specific rates.  The data will be available to States later this year …”

The Secretary’s decision to rely on a survey of drug acquisition costs is somewhat ironic; by law manufacturers are required to disclose average manufacturer price (AMP) to the Secretary, though the Secretary’s ability to share AMP data has been limited by a string of federal laws and judicial decisions[2]. Also, the definition of AMP has been shifted by legislation over time, and the proximity of AMP to true acquisition cost has likewise shifted as a result. We speculate that the Secretary’s decision to rely on surveys of actual acquisition costs is an attempt to circumvent both of these problems – acquisition cost data obtained by survey would not be subject to laws or rulings that prohibit the Secretary from making public AMP data provided by manufacturers; and, the Secretary would be free to define acquisition costs however she chooses, and would presumably choose a definition that avoids as many of the artifacts associated with AMP as possible

This leads to the question of whether the acquisition cost data will be available to sponsors of private plans – and we presume they will. Alabama was the first state to move to a survey of actual acquisition costs, and in the process has fought many of the judicial battles that attend such a move. Alabama’s decision to publish their acquisition cost data appears to have survived these challenges, so to the extent that Alabama’s judicial experience predicts the outcome of challenges to HHS actions, the HHS Secretary also should be able to make acquisition cost data publicly available. And, given the current political climate surrounding health cost growth and the contribution such data would make to lowering health costs, we presume the HHS Secretary, as a member of the President’s cabinet, would have motive to make the data public

Bearing in mind that AWP is a benchmark that was brought into being by Medicaid / Medicare, the shift by CMS to actual acquisition costs would obviate the need for AWP in government re-imbursement. Without government support we believe AWP disappears; it becomes meaningless in the context of government re-imbursement, and sponsors of private / commercial plans should plainly prefer actual acquisition costs over AWP, as AWP plainly works to their dis-advantage

Recall that pharmacies earn roughly $5 more per generic prescription dispensed than per brand prescription, and that we’ve argued this incremental margin is attributable to some combination of two factors: plan sponsors’ interest in tilting the channel in favor of generics; and, plan sponsors’ inability to see pharmacies’ generic acquisition costs. As regards the former, we’ve argued that sponsors’ need to bias the channel in favor of generic dispensing is an inverse function of beneficiaries’ preference for generics over brands. In the past, beneficiaries have tended to prefer brands over generics, in large part because generic v. brand co-pay differentials were too narrow to overcome beneficiaries’ natural ‘all else equal’ tendency to prefer brands. As generic v. brand co-pay differentials widen, beneficiaries increasingly prefer generics, which eliminates the need for plan sponsors to offer the trades higher margins on generics, since the drug trades, like any other distributor, as a practical matter must stock what consumers prefer

As regards the latter factor driving higher generic dispensing margins, namely sponsors’ inability to see generic acquisition costs, we’ve argued that because the relationship between the prevailing benchmark price (AWP) and brand acquisition costs is highly consistent (AWP = 1.2x brand acquisition cost), the common “AWP – X%” formula used for re-imbursing pharmacies ‘works’ in the (increasingly narrow) case of brand re-imbursement. E.g. because brand AWP’s are consistently 1.2 times brand prices, contract terms re-imbursing pharmacies at (for example) AWP – 12.5% for brand dispensing would consistently produce a 5% dispensing margin on brands[3]. Because the AWP – X% formula produces a consistent mark-up in the case of brands, the ‘value’ of this mark-up represents an efficient pricing of the brand dispensing ‘service’ between fully informed buyers and sellers

However in the case of generics, the relationship between AWP and actual acquisition costs is highly varied, and more or less evenly distributed around the mean. Thus the common “AWP – X%” formula is very nearly useless to plan sponsors; because the relationship between AWP and pharmacies’ true generic acquisition costs is random, so is the relationship between generic acquisition costs and AWP minus any fixed percentage. If sponsors set AWP – X% such that pharmacies earn just enough on the average generic to make dispensing worthwhile (i.e. a competitively efficient margin, as exists in brand dispensing), then pharmacies would not earn sufficient dispensing margin on half of the sponsors’ beneficiaries’ prescriptions. Accordingly sponsors are forced to set AWP – X% such that pharmacies earn much more than a competitively efficient dispensing margin on the average generic prescription. We’ve compared this to flying over mountains in a cloud bank with nothing but an altimeter – the relationship between your altitude and the average height of peaks in the range is of no interest; all you really care about is the spread between your altitude and the height of the tallest peak in the range. A shifting of the benchmark to acquisition cost would eliminate this uncertainty, with the likely result being compression of generic dispensing mark-ups to a competitively efficient level. As this is clearly in the best interest of plan sponsors, and as plan sponsors have sufficient leverage over the trades to specify the benchmark they prefer, we believe that commercial PBM contracts will be shifted from AWP to acquisition costs, and that generic dispensing margins will compress as a result

US drug pricing tends to decelerate in presidential election years; drug manufacturers and wholesalers are most susceptible

In our last call, we outlined three conditions that we believe set the stage for rapid rates of list price growth for US drugs:

  1. a common perception that immediate risks of incremental legislative / regulatory restrictions on pricing are low;
  1. a common perception that pricing freedoms eventually will be lost, thus creating motive to raise prices rapidly while this remains possible; and,
  1. a widespread need for real pricing gains to offset fundamental weaknesses elsewhere

All three conditions presently are met, and US real drug list prices have grown by roughly 6.5% since the 2008 election. With a general election approaching in 2012, the first of these three conditions arguably is lost, especially in the likely event that the upcoming general election features healthcare among its major campaign themes

To help frame the relevance of the political cycle to US drug pricing, we examined real rates of US prescription drug list price growth since 1972 (Exhibit 3); presidential election years are shown in bold. Taking the results in aggregate, presidential election years tend to be characterized by decelerating real rates of US drug pricing growth (Exhibit 4) both in absolute terms, and also relative to either mid-term election years or the years immediately before or after elections. In five of the last ten presidential elections drug pricing grew in real terms the year before the election – as is the case presently — and in all five instances drug pricing decelerated during the subsequent election year (Exhibit 3, again)

Two sub-sectors in particular would be affected by a decelerating real pricing trend: manufacturers and drug wholesalers. The logic surrounding manufacturers is obvious and straightforward, though we would emphasize that manufacturers’ level of reliance on real pricing growth is perhaps greater than it’s ever been. Exhibit 5 shows the aggregate rate of US sales growth for large cap manufacturers since 1983, and the ratio of pharmaceutical PPI to sales growth. Sales growth is at a trough, but reliance on real pricing power is at a peak

This argues that a deceleration in the real rate of US drug pricing could drop aggregate US sales growth well into negative territory, even before considering the effects of pending patent losses. On the assumption that manufacturers take fairly consistent advantage of the prevailing real price trend, i.e. that weighted average real pricing across each manufacturer’s product line is roughly equal to the prevailing average, relative susceptibility to a decelerating real price trend is a function of each manufacturers rate of US sales growth – i.e. the slower the company’s US sales growth, the larger pricing gains are as a percentage of that sales growth. Exhibit 6 ranks the large cap manufacturers according to immediate past rates of US sales growth

Wholesalers’ buying margins are affected by the pace of real pricing; before the widespread use of inventory management agreements (IMA’s), wholesalers routinely speculated on pending pricing actions by carrying higher inventories of products that were expected to be priced most aggressively. Now, with IMA’s commonplace, wholesalers have limited

ability to speculate on pricing by loading in inventories; instead, wholesalers’ rate of participation in product price increases is generally fixed. We believe that most agreements still allow wholesalers the benefit of pricing actions, though on a pre-determined level of inventory

Before IMA’s, the opportunity set for inventory investments was the net of pharmaceutical PPI growth over the costs of carrying incremental inventory, or roughly pharmaceutical PPI – LIBOR plus a small premium. As a backdrop for our arguments, Exhibit 7 shows the history of pharmaceutical PPI – LIBOR over the last decade. IMA’s became the norm by the mid-2000’s, so wholesalers’ exposure to real price gains since this period generally has been fixed by terms of these agreements; and notably, under IMA’s wholesalers generally are not required to load in inventories to gain the benefit of manufacturers’ pricing actions – thus wholesalers no longer face significant marginal short-term borrowing costs as an offset to real price gains on inventory. Thus despite the restrictions on speculation posed by IMA’s, we suspect the contribution of real price gains to wholesale buying margins is non-trivial, and that a sudden deceleration of real pricing, as we anticipate in early 2012, could pressure wholesale margins

An expanded view of drug stock selection

In previous work[4] we’ve shown that drug stocks with proximate[5] major product approvals tend to outperform their peers, and that with the exception of the period immediately surrounding the anticipated regulatory decision[6], that relative performance gains are fairly steady well before and well after product approval. The consistency and magnitude of relative performance gains, and their tendency to be spread fairly evenly across time, generally supports the notion of favoring drug stocks with pending (or recent) approvals, and specifically argues against the notion that betting on product approvals is a high-turnover strategy

Because we expect the industry to lose its ability to raise pricing in real terms, and because we see continued consumer elasticity headwinds as co-pay and co-insurance rates rise, we heavily weight the only remaining driver of revenue growth – product mix – as an additional basis for stock selection. We capture product mix by calculating the sales-weighted average age of each company’s marketed product portfolio, the idea being that younger portfolios generally are more likely to be on the upswing of the product lifecycle

And finally, because of the enormous effect of patent losses on returns to R&D spending, and because of our belief that unlike small molecules, large molecule innovators will retain substantial market share at reasonable prices even after the entry of follow-on biologics[7], we greatly prefer companies with relatively high shares of sales attributable to biologics

Exhibit 8 sorts the US-listed drug manufacturers according to these three criteria; this list builds on our previous screen of large cap manufacturers only, adding specialty and biotech names

To be clear we view such screens more as starting points than as destinations; and, we tend to view this screen in particular as better able to rule names out than to rule names in. Even without penalizing companies for large size, AZN, PFE, GSK, NVS, and JNJ rank near the bottom on these key metrics. Because of their large size these names tend to be driven by systematic risks, virtually all of which we view as negative; and, lacking any of our three favored characteristics (proximate major product approvals, relatively young product portfolios, and substantial exposure to biologics), all of which are geared to select for beneficial idiosyncratic risks, we would tend to avoid these names

  1. We use a discount rate of 9.39%, though the relative value conclusions of this analysis are not terribly sensitive to the discount rate used. In the case of the zero R&D scenario, we assume an annual sales erosion rate of 5.25%, which is equivalent to the long-term average of this year’s sales losing sales next year for a typical small molecule portfolio
  2. To date, efforts to replace AWP have been a series of false starts. The Deficit Reduction Act (DRA) of 2006 called for CMS to set generic re-imbursement off of average manufacturer price (AMP), and to make AMP publicly available. The National Association of Chain Drug Stores (NACDS) sued CMS to block this action, and won a preliminary injunction from the US District Court for the District of Columbia in December of 2007, and this injunction remains in effect, though it appears to be mooted by the passage last year of the Patient Protection and Affordable Care Act (PPACA). PPACA instructs the Secretary of HHS to make AMP publicly available, and to set generic re-imbursement as a function of AMP, and calls for this to take place by January of 2011. The Government Accountability Office (GAO) on December 15 of last year disclosed in a letter to Congressman Waxman that as of October 2010, HHS had not taken steps to comply with these PPACA provisions, and January 2011 obviously has come and gone without action by HHS
  3. 1.2 – (1.2 x 0.125) = 1.05
  4. September 22, 2010, “The Skeptic’s Guide to Drug Stock Selection”
  5. Either occurring within the next two years, or having occurred within the past two years
  6. Risks are asymmetric roughly 90 days preceding scheduled US regulatory actions; rates of relative performance are no greater here, but presumably because bad news tends to precede scheduled regulatory actions, downside risks are very concentrated in this period
  7. January 26, 2011, “Ranked Preferences Across Healthcare, Subsector by Subsector”
Print Friendly, PDF & Email