ESRX, MHS, and the PBM Bear Case

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

Richard Evans / Scott Hinds

203.901.1631 /.1632

richard@ /

July 25, 2011

ESRX, MHS, and the PBM Bear Case

  • We update and summarize our PBM bear case in light of ESRX’ bid for MHS
  • Generic dispensing margins (est. $9.01 / Rx) should fall toward brand levels (est. $5.77 / Rx) if average wholesale price (AWP) is replaced by a benchmark that more closely reflects true generic acquisition costs. Two candidate benchmarks are emerging: average acquisition cost (AAC, essentially a survey of retail prices), and average manufacturer price (AMP, essentially a more reasonable proxy for manufacturers’ net selling price). AAC should be nationally available as soon as late this year, AMP by the end of 2012
  • As mix shifts away from highly interchangeable small molecule brands, fewer rebate dollars are available. We estimate that three-quarters of dollar sales attributable to highly interchangeable small molecule brands will be lost to generics within three years
  • As mix shifts toward specialty products, management of the drug benefit tends to be reintegrated into the medical benefit, i.e. taken away from PBMs. Patients taking specialty brands tend to be more closely and actively managed by medical plans, thus favoring reintegration of the drug and medical benefits. We suspect this at least partially explains UNH’s decision not to renew the MHS contract
  • Because of the two preceding points, the large PBMs’ infrastructure (call centers, mail-order) is less essential, thus the large PBMs are a less and less cohesive oligopoly
  • The emergence of co-pay cards further limits PBMs’ ability to extract large brand rebates from manufacturers, especially for specialty drugs; this in turn further limits PBMs’ ability to offer an (absolutely or relatively) attractive value proposition in specialty pharmacy management
  • The declining relevance of call center and mail-order infrastructure, the increasing relevance of generic dispensing spreads to total potential savings, and the ability of smaller ‘b-list’ PBMs to offer tight generically oriented formularies, all point to share gains by b-list PBMs, even as the total market for PBM services may narrow
  • The high likelihood of employers shifting subsidy-eligible employees out of employer sponsored insurance (ESI), and onto health insurance exchanges (HIEs), means potential PBM beneficiaries will be shifted from employers (where PBMs have relatively high odds of capturing beneficiaries at relatively healthy margins) to HMOs (where both the odds of capturing beneficiaries and service margins are far lower)
  • We recognize that consolidation of ESRX and MHS would tend to lessen the tendency for price competition among the a-list PBMs; however, we would emphasize that CVS is still motivated to price compete because it enjoys marginal retail volume gains that ESRX and MHS do not; and, that the majority of threats to the large PBMs tie to issues other than price competition
  • The prospect of an ESRX/MHS combination does not change our view of the PBM industry, nor would we view ESRX/MHS together as significantly more able to react to these changes than either PBM on its own

Since March of 2010 we’ve published a series of research notes on the PBM industry; in light of ESRX’ bid for MHS, this note updates and summarizes our views

Broadly, our concerns fall into two categories: threats to the relatively higher dispensing margins earned on generics vs. brands; and, other threats to overall pricing and profitability

Relative Profitability of Generic Dispensing

Our most pressing concern centers on the higher mark-up PBMs receive on generics than brands. We believe higher relative margins on generics are an artifact of the current standard for commercial contract benchmarks, namely average wholesale price or AWP; and, we believe that if and when AWP is replaced, that absolute margins per generic prescription dispensed (which we estimate to be roughly $9.01) will fall to or at least very much toward the level we estimate for brand prescriptions ($5.77). We believe that an alternative to AWP will be available as early as late 2011[1], and that either of two potential alternatives might begin finding its way into commercial contracts as early as late 2012 / early 2013. The two alternative benchmarks are average acquisition cost[2] (AAC), and average manufacturer price[3] (AMP). Health and Human Services (HHS) Secretary Sebelius promised the states a federal version of AAC by late 2011 in an open letter published on February 3rdof this year[4]; and, the Affordable Care Act calls on the Secretary to publish AMP[5] in 2011, though publication appears unlikely before 2012

AWP is 1.2x wholesale acquisition cost (WAC). For brands, WAC is roughly equivalent to the actual cost to a wholesaler (or warehousing retailer, or PBM) of acquiring a brand, since manufacturers do not discount brand drugs to the ‘trades’ (wholesalers, retailers, and PBMs). Therefore AWP is roughly equal to 1.2x the trades’ acquisition cost for a brand (but not for a generic, more on that in a moment). Commercial contracts generally call for the trades to be reimbursed for brand prescriptions at AWP minus some fixed percentage; and, since AWP bears a consistent relationship to true brand acquisition cost (~1.2x), commercial reimbursement formulas using an ‘AWP-minus’ basis are in effect very straightforward ‘cost-plus’ arrangements for brands. Thus pharmacies and PBMs earn fairly constant percentage mark-ups on each brand dispensed, the only differences being that more expensive brands obviously yield higher absolute margins, and various payors reimburse at various discounts to AWP (or, in effect, mark-up to brand acquisition cost). Exhibit 1 shows the absolute dollar dispensing margins for brands dispensed to Medicaid beneficiaries in Alabama in 1Q/2011; margins are ranked from lowest to highest on the x-axis. Percentage mark-ups are constant, absolute dollar mark-ups vary only as a function of the acquisition cost of the drug, and no brand drugs have a negative dispensing margin, since the ‘plus’ in the effective cost-plus formula is in fact always positive

Generic reimbursement works off of the same basic formula, but the result is quite different. AWP is still 1.2x WAC, but WAC (effectively the list price of the generic) bears little or no relation to the trades’ actual costs of acquiring the generic, since generic manufacturers sell to the trades at (extremely varied across molecule, manufacturer, and time) discounts to list. Thus AWP bears very little relationship to true generic acquisition cost; and by extension, AWP-minus reimbursement formulas result in extremely varied dispensing margins across generic products. Exhibit 2 shows the highly varied relationship between dispensing margin and acquisition cost for generics; and for comparison, the straight-line relationship between dispensing margin and acquisition cost for brands. Because of this varied relationship unlike what we see for brands, AWP-based reimbursement formulas do not function in a cost-plus manner for generics. Exhibit 3 is the ‘generic version’ of Exhibit 1, where we graph absolute dollar dispensing margin by (in this case generic) drug from lowest to highest. At present generic reimbursement levels of roughly AWP-50% several drugs actually have negative dispensing margins, most drugs have very small dispensing margins, and a modest number have very large dispensing margins

If a payor insists on a larger discount to AWP for brands, the trades’ brand dispensing margins of course fall, but they can never fall below zero[6]. In contrast, if a payor insists on a larger discount to AWP for generics, the result is that all drugs make less money, and a larger proportion of drugs lose money. Exhibit 3 illustrates this effect by showing what happens to dispensing margins (ranked from smallest to largest) as payors insist on ever-larger discounts to AWP

Beyond the obvious fact that greater generic discounts to AWP reduce average generic dispensing margins, these greater discounts create two further problems: 1) as a larger proportion of the generics lose money, payors face a growing risk that the trades will not stock essential generics; and 2) because the few drugs that make very large margins for the trades only do so for a short period of time, the trades face a growing risk that average generic dispensing margins become very small during periods when few new generics are being launched. We believe that because both the payors and the trades face real consequences as the ‘minus’ in the AWP-minus generic reimbursement formula grows larger, that the current ‘minus’ level is in fact relatively stable. More simply, as long as AWP is the benchmark for reimbursing the trades, generics will carry higher absolute margins than brands

However, if AWP is replaced with anything that is more indicative of actual generic acquisition costs, we would expect the generic reimbursement ‘dynamic’ to shift to cost-plus, in which case we see very little reason for generic dispensing margins to be (much if any) higher than brand dispensing margins[7]. Both of the anticipated new benchmarks, AAC and AMP, are reasonably reflective of true generic costs. If either is available, it is to payors’ advantage to replace AWP[8], and shift payment for generics to a cost plus (i.e. AAC-plus or AMP-plus) basis – which we fully believe commercial payors can and will insist upon

The Shift in Mix from Brands to Generics, and from Small to Large Molecules

The origins of the PBM industry lie in a period characterized more by brands than generics, and more by ‘traditional’ small molecules than by specialty (often large molecule) drugs. The industry’s first act was to negotiate rebates on highly interchangeable small molecule brands, dispensed at retail to patient groups that were largely homogenous, and whose disease states generally were neither critical nor complex. Early examples included ACE inhibitors, NSAIDs, H2 antagonists, and non-sedating antihistamines; more recent examples include angiotensin-II receptor antagonists (ARBs) and statins. Formulary managers generally and PBMs particularly could play brands off one another in each category, the result being large rebates paid by these brands for access to restrictive formularies

Extracting these brand rebates meant having the necessary infrastructure to make good on the threat to exclude a given brand – in particular, large call centers and mail-order warehouses. A significant percentage of covered patients arrive at retail pharmacies with prescriptions for non-preferred brands; large and efficient call centers allow formulary managers to quickly attempt having the prescription switched to the preferred brand in the very few minutes available before the patient leaves. Mail-order facilities offer the dual benefits of 1) slowing down the prescription fill process so that call centers have more time to contact physicians and switch prescriptions to preferred brands, and 2) capturing a larger percentage of the available margin, especially in the case of newer generics

The result is that tightly managed formularies offer payors lower cost access to brands – as of 2009, we estimate that brands in competitive categories were offering rebates in the neighborhood of 40%. Thus as long as the prescription market consists in large part of highly interchangeable brands that can be switched at retail, payors are far better off having their drug benefits ‘actively’ managed, e.g. through a PBM, than they are purchasing drugs at retail prices

However, as mix shifts from highly interchangeable brands to generics, there is less need for PBMs’ traditional call center and mail-order services. Of the dollar sales coming from highly interchangeable brands in today’s market, we believe brands representing three-quarters of these dollar sales will be available generically in three years. And, because new product flow is not keeping pace with patent losses at an industry level, and also because an increasing proportion of new product flow falls into less interchangeable specialty categories, we believe highly interchangeable brands become a very small proportion of drug sales within these next three years

This shift away from highly interchangeable brands has two effects – it mitigates the need for massive call center and mail-order infrastructure; and, it forces PBMs to offer an alternative value proposition. As regards the need for call centers and mail-order: if the task is to get the patient on the generic version of the brand, this can be done most efficiently with a simple hard-code: e.g. Lipitor is not covered, but generic atorvastatin is. Any PBM can offer this type of hard-code, including PBMs without large call centers or mail-order facilities

At least in theory, an alternative value proposition might come in either or both of two forms: lower generic dispensing margins, and/or more active management of (and larger rebates from) large molecule / specialty drugs. As small molecule generics increasingly dominate retail prescription volumes, with generic dispensing margins being a great deal (+/- $3.24) higher than brand dispensing margins, and with smaller PBMs increasingly able to offer services as the need for large call centers and mail-order facilities becomes less acute, competition amongst PBMs to lower generic dispensing margins is a very likely outcome – but if and only if AWP is replaced. In fact the scale of savings payors might enjoy from lower generic dispensing margins is not wildly different from the scale of savings produced by brand rebates during PBMs’ recent history. From 2002 to 2009 we estimate the large PBMs produced an average brand rebate of roughly $4.75, and that brands were half of prescriptions dispensed (weighted average = $2.38 / Rx across all Rx’s). In 2011, reduced generic dispensing margins can offer as much as $3.24 in savings across more than 70 percent of prescriptions dispensed (weighted average = $2.27 across all Rx’s)

Obviously reducing generic dispensing margins means reducing the PBMs’ own margins, thus the preferred value proposition – in theory — would be to produce from large molecule / specialty drugs the types of rebates the larger PBMs captured from small molecule / retail drugs. In practice, we don’t believe this is feasible, for two reasons: first, specialty products are less interchangeable; second, co-pay cards are reducing PBMs’ rebate leverage

In the ‘old’ brand rebate scenario a 40-year old with a first-time Lipitor prescription presents himself to a retail pharmacy. His formulary prefers Crestor to Lipitor, so as the patient waits for his first-ever statin prescription his PBM’s call center attempts to contact the physician and switch the prescription to the preferred brand. Essential elements of this story are that: 1) the odds of the health plan knowing ahead of time that the patient would be prescribed a statin, or if they did know, to intervene, are low; 2) if a switch is going to happen at retail it has to happen very quickly; 3) the patient is not critically ill and his condition is unlikely to be at all different from that of other patients with the same diagnosis, so for all intents and purposes either drug will do; 4) the patient’s physician does not have so strong a preference for Lipitor in this case that she will fight to have Lipitor reimbursed; and 5) the health plan’s medical director, recognizing that Crestor and Lipitor are largely interchangeable, has agreed to allow the formulary manager to choose which brand is preferred for reasons more economic than clinical. On net, the formulary manager is in a position to routinely exclude Lipitor (or if it hadn’t made the deal, Crestor), so the rebate Crestor had to pay to win its preferred position is a large percentage of its retail price

In the ‘new’ brand rebate scenario a 50-year old presents at pharmacy with a first-time prescription for a biologic (e.g. Enbrel) to treat his rheumatoid arthritis (RA). In this case all of the essential elements of the preceding story are reversed: 1) the health plan is very likely to have known that this specific patient has RA, and is very likely to have actively managed this patient’s care before he presents to the pharmacy; 2) early intervention in drug selection for this patient is not confined to the few minutes he might spend at retail; 3) the patient is quite ill and his condition is quite likely to differ from that of other patients with the same diagnosis, so drug selection is clinically important; 4) in light of the preceding, the physician is reasonably likely to have a strong preference for one drug over another; and 5) recognizing that product selection in this category is clinically important, and that several drug options may be necessary to effectively treat the plan’s broader population of patients, the plan’s medical director weights clinical considerations at least as heavily, and perhaps more heavily, than economic considerations in developing the formulary, and so exerts more influence over formulary design than was the case with statins

In the first example, the PBM has high degrees of autonomy (from the medical director, who recognizes the drugs are highly similar and allows the drug decision to be driven largely by economics) and rebate leverage (over the manufacturers, who recognize that not paying a sufficient rebate means being largely excluded from formulary). And, the PBM benefits from being uniquely capable in this scenario, as its call center and mail-order facilities are essential to the overall process. In the second example, the PBM has less autonomy (the medical director prioritizes clinical considerations at least as highly, if not more highly, than economic considerations), less rebate leverage (manufacturers recognize that the worst case scenario is being down a step or two in the rank order, rather than being excluded), and is less uniquely capable (its call centers and mail-order facilities are useful, but not essential). Thus as retail brand mix shifts from ‘traditional’ small molecules to ‘specialty’ large molecules, we see the large PBMs as less critical to the overall process of managing formularies and negotiating rebates, and so believe the PBMs naturally come under price / volume pressure. In fact, we believe that because specialty drug selection is so closely integrated into the broader clinical management of the patient, and that because medical plans (having full view of the medical record and an existing ability to intervene across the care continuum) are so much better positioned to manage these patients, that as drug mix shifts toward specialty, that the prescription benefit will frequently be reintegrated into the medical benefit. UNH’s recent decision not to renew with MHS, and presumably to manage its own drug benefits, is at least consistent with our interpretation

Co-Pay Cards as a Further Limit on Drug Rebates

The emergence of co-pay cards further reduces formulary managers’ rebate leverage over brands – irrespective of whether the drug benefit is managed as a carve-out by a PBM, or as an integrated part of the drug benefit by the medical plan. As brand drug rebate percentages have grown, and as brand drug retail prices have grown, the absolute dollar value of rebates for those drugs offering rebates has reached a level that is roughly equal to patients’ marginal co-pays. Assume for example a 3-tier formulary with co-pays of $10 for generics, $30 for preferred brands, and $50 for non-preferred brands. A brand with a retail price of $150 might have to offer a 15 percent ($22.50) rebate to be on the non-preferred 3rd tier, with a $50 co-pay. Before co-pay cards, manufacturers would offer whatever rebate was efficient in order to get to the $30 co-pay preferred tier. For example the manufacturer might decide that the volume / price trade off of giving another 20 percent ($30) rebate, on top of the base $22.50 rebate (total $52.50), would be an efficient choice. However with co-pay cards, the manufacturer now has the option of paying the $22.50 ‘base’ rebate for tier 3 access, and offering patients co-pay cards to offset the patient’s $20 marginal difference in co-pays between tiers 3 ($50 co-pay) and 2 ($30 co-pay). Ignoring frictional costs of the co-pay card, the manufacturer now has ‘effective’ tier 2 status, for a total discount from retail of ‘only’ $42.50. The simple point is that manufacturers will tend not to offer marginal rebates for preferred formulary status that are greater than the difference between co-pays for non-preferred and preferred tiers. This is a relatively new constraint on rebate growth, for three reasons: 1) ‘efficient’ co-pay cards are themselves new; 2) because of larger percentage rebates and larger retail prices we have only recently reached a point at which marginal rebates for preferred access are as large as the difference between non-preferred and preferred co-pays; and 3) specialty drugs tend to have very high retail prices in relation to co-pay levels. We believe formulary managers cannot raise co-pays fast enough to return to past levels of absolute dollar growth in rebates – co-pay levels already are at a point where consumers forego medically necessary drugs, and plan sponsors are hesitant to allow co-pays to rise quickly, if at all. The co-pay card constraint on rebates is particularly relevant in the case of specialty drugs, where retail prices are very high relative to rebates. Rebates earned on these drugs are never likely to be as large in percentage terms as those earned on traditional brands, where co-pays are a higher percentage of retail price. Thus because of co-pay cards formulary managers are less able to deliver a given percent reduction in specialty brand costs, and this constraint on negotiating leverage is in addition to the other considerations we’ve raised regarding formulary managers’ relative inability to exclude specialty drugs from formulary

Summarizing our PBM Thesis, and What ESRX/MHS Means in This Context

Generic dispensing margins are an artifact of AWP-based reimbursement; these should fall to or toward brand levels if and when alternatives to AWP are available. We believe AAC will be available in late 2011 / early 2012, and AMP soon thereafter. Once either or both of these benchmarks are trustworthy, i.e. have survived legal challenges, we expect commercial payors to insist on either benchmark, displacing AWP. We would generally expect either alternative to begin appearing in commercial contracts as early as late 2012 / early 2013. Given the savings to sponsors, we would expect many contracts to shift before their schedule expirations

PBMs are less and less able to deliver large brand rebates, which is the historic foundation of their value proposition, for 3 reasons: 1) highly interchangeable (e.g. small molecule retail) brands are a falling percentage of total drug spending; 2) newer (large molecule specialty) brands are far less interchangeable, and thus do not have to offer very high percentage rebates for formulary access; and 3) rebate levels are effectively capped at prevailing preferred v. non-preferred co-pay levels

Plan sponsors increasingly have alternatives to the large PBMs: 1) as brand mix shifts toward specialty products, management of the drug benefit will tend to be reintegrated with the medical benefit; i.e. health plans’ captive PBMs have an advantage over stand-alone PBMs, which is a reversal of the status quo; 2) the large PBMs’ call center and mail-order infrastructure, essential for extracting rebates from interchangeable small molecules at retail, is far less essential as highly interchangeable retail brands fade as a percent of drug benefit costs; this erodes the foundation of the large PBMs’ effective oligopoly; and 3) as generic prescriptions dominate retail volumes, pressure on generic dispensing margins will increase; because brand to generic shifts can be managed without traditional large PBM infrastructure (call centers, mail-order), b-list PBMs are increasingly attractive alternatives to the larger PBMs

We believe health reforms will result in a shift of beneficiaries from employer-sponsored insurance (ESI) to the health insurance exchanges (HIEs). This results in a shift of potential PBM beneficiaries from employers (where PBMs have had relatively high odds of capturing business at relatively healthy margins) to HMOs (where PBMs have lower odds of capturing beneficiaries and where margins are lower)

An essential moving part to our PBM thesis is that the large ‘a-list’ PBMs (ESRX, MHS, CVS) are a fading oligopoly. Inherent in this argument is an expectation that pricing competition heats up among the a-list players, and recent history offers plenty of evidence that this is happening. An ESRX/MHS merger plainly should reduce the tendency of this legacy oligopoly to tip toward price competition. Nevertheless we still see mounting price competition among the a-list players, even if the list is reduced to two. Most importantly, the basis of the a-list’s competitive advantage (the historically essential nature of call centers and mail-order) continues to fade, so as we’ve emphasized above plan sponsors have more options for managing their drug benefits. And, because CVS gains more from PBM price competition than either ESRX or MHS (CVS gains not only the PBM volume, but some retail volume as well), separate or combined, we believe CVS’ relatively greater tendency to reduce its PBM pricing will continue to be a key driver of the a-list pricing trend

Whether pricing competition heats or cools within a narrowed a-list, we’re still left with a very long list of fundamental threats to the a-list that should unfold whether or not an ESRX/MHS merger occurs – accordingly the merger would not affect our broader PBM thesis, nor would we view an ESRX/MHS NEWCO as insulated from these threats

  1. If either alternative benchmark comes by late 2011 it would almost certainly be AAC
  2. AAC is a survey-based calculation of the average cost of retail pharmacy acquiring a given drug
  3. AMP is a value reported by manufacturers to HHS; this value is an imperfect estimate of the net price at which generic drugs are sold to pharmacies, though it is a far more representative value than AWP
  5. (pgs 211 & 212)
  6. As long as the ‘minus’ in AWP-minus is less than 20 percent
  7. We acknowledge that in the past, it has been necessary for the trades to earn more on generics than on brands, simply in order to ensure that generics were dispensed instead of brands. This was true because patients did not prefer generics to brands, and at then-prevailing co-pays often preferred brands to generics. As co-pay differentials between brands and generics become far larger, consumers’ preference for generics over brands grows, and this reduces the need for trades to earn significantly more on generics than brands
  8. We acknowledge that many generic drugs are reimbursed at maximum allowable cost, or ‘MAC’, instead of AWP. Our calculations of generic dispensing margins take this into account; where drugs have a federal upper limit (FUL), we assume the generic is ‘MAC’d’. MAC prices are better, but still very poor, approximations of true acquisition cost than AWP. Our work shows that dispensing margins on MAC’d generics also fall over time, i.e. MAC prices slowly ‘chase’ true acquisition cost. With either AAC or AMP, reimbursement would very quickly assume a fairly tight relationship with true acquisition cost
Print Friendly