2010 HMO and PBM GM’s Expand; The Public Plan Has the Blues
November 2, 2009
2010 HMO and PBM GM’s Expand; The Public Plan Has the Blues
Claims Cost Pressure Should Ease in 2010, Aiding Insurer’s Margins
- We show that fear of job loss drives employees to accelerate healthcare consumption; and, in periods of substantial job loss, that this effect is >= 2x larger than the effect of COBRA enrollment on insurers’ claim costs.
• As job losses abate, so do the related pressures on medical loss ratios (MLR’s) – even if unemployment remains high. Gross job losses are moderating; this ‘08/’09 pressure on MLR’s should be largely gone in ’10.
• We estimate the fear-of-job loss effect over the ’01-’03 downturn; the effect added roughly 3.2% to MLR’s. Given much larger job losses in ‘08/’09, the effect almost certainly is now much larger; it’s absence in ’10 augurs for lower MLR’s, and thus higher gross profits for insurers.
Why Think of the Public Option as Anything Other Than a Blues Plan?
• Both chambers of Congress now include public options in their reform bills; neither give the public option access to Medicare payment rates. It appears the public option would be required to negotiate price with providers, pay back its start-up capital, cover its operating costs, and earn sufficient reserves. In other words, it looks like a Blues plan.
• We compare measures of competition / market concentration in states with for-profit Blues to states with non-profit Blues. On average, non-profit Blues have 4 percent greater share than the for-profit Blues, and market concentration measures are slightly higher (HHI of 3779 in non-profit states, v. 3319 in for-profit states). Large non-profits have a measurable advantage, but the advantage is relatively small.
Drug (List) Prices Rise; Drug Discounts to Commercial Buyers Deepen; PBM’s Make Money Both Ways
• We see high odds of accelerated pharmaceuticals list pricing in 2010. Immediate motives include pending patent expirations, greater Medicaid rebates, and a new tax. Longer-term motive includes the considerable loss of pricing freedom if a Medicare Commission passes into law. As an aside, we see this as the single most fundamental risk to innovators’ margins from pending legislation, but sense its relevance is not fully recognized by the capital markets.
- Opportunity takes the form of political distraction; after the reform debate ends – whatever its outcome – we do not see Congress giving attention to healthcare in 2010 (and perhaps a few years beyond), even for the purpose of disciplining pharmaceuticals manufacturers for accelerated list pricing.
• Despite list price growth, net gains to manufacturers are modest. We expect deeper commercial discounts as the minimum Medicaid rebate grows from 15.1% to 23.1% in 2010. Manufacturers have long stopped short of giving commercial discounts beyond the 15.1% threshold, to avoid the requirement of having to give Medicaid the same pricing. As this threshold falls to 23.1%, this source of net pricing discipline fades, thus our expectation of deeper commercial discounts.
• PBMs are clear winners, being positively levered to: accelerated real pricing, deeper discounts, and (brand pharmaceuticals) patent expiration. If real pricing grows faster and discounts fall further than expected, PBMs have the potential for positive earnings surprises. Wholesalers also benefit from accelerated real pricing and generic competition, but gain nothing from deeper commercial discounts.
Higher than expected claims costs tie to fear of job loss; this effect is larger than COBRA, and should now have dissipated, alleviating MLR pressures into 4Q/’09.
The large-capitalization health insurers’ business lines are heavily concentrated in employer-sponsored insurance (ESI) markets, making loss of enrollment inevitable in economic downturns. The tendency of some portion of the newly unemployed to seek extended coverage through COBRA is well understood, as is the fact that because of much larger out-of-pocket premium costs, COBRA enrollees are highly self-selected – i.e. sicker — and so tend to have greater claims costs than the average employee. Thus not only do insurers suffer dis-enrollment during economic downturns, claims costs rise relative to premiums as COBRA enrollees represent a larger proportion of the enrollees that remain.
Insurers’ results began to show the effects of a weakening economy as early as spring of ’08; and, as the economy has worsened, the magnitude of claims cost pressure has been greater than would have been expected due to the COBRA effect alone. Understandably this raises a classic cyclical fear, namely that of declining secular pricing power, with the attendant risks of earnings pressure over a multi-year period.
We see few if any of the underpinnings of a secular decline in pricing power, and believe the majority of current MLR pressures tie to an alternative explanation – that sharp declines in employment create significant fears of job loss across a significant portion of the still-employed population; and, that employees having a high index of job-loss fear accelerate their consumption of health care while they still have ESI.
Using Medical Expenditure Panel Survey (MEPS) data from 2001 – 2006, we compared the economic (and employment) decline of ’01 – ’03 to the recovery of ’04 – ’06. We made the assumption that employees, on average, have some reasonable idea of whether they are likely to lose their job. By extension, we reasoned that employees who actually lost their jobs, i.e. who in fact had the greatest risk of job loss — since it happened – were more fearful of job loss before the fact than employees who kept their jobs.
Across the ’01 – ’06 timeframe, we found that persons losing their jobs had health claims that were 41% higher than persons who kept their jobs (Exhibit 1, upper left solid arrow). Note that the relatively greater claims of those laid off is not due to greater age, in fact those laid off were on average 4.9 years younger than those that kept their jobs (also Exhibit 1). Building on the notion that those losing their jobs tend to see it coming, we would expect this fear level to be greater during downturns, and smaller during upturns. By extension, we would expect the excess claims of those that lost their jobs to be larger during the ’01 – ’03 downturn (Exhibit 1, upper middle arrow), than during the upturn (Exhibit 1, upper right arrow). This is in fact the case; the gap between those who lost or kept their jobs is larger in ’01 – ’03 than in ’04 – ’06. We find the evidence compelling, and so believe that fear of job loss leads at-risk employees to accelerate their health spending while still employed.
This leaves the question of which effect – COBRA or fear or job loss – has the greater impact on MLR’s. We also examined COBRA-enrollees’ claims costs, and compared these to persons that remained employed. Across the ’01 – ’06 period, COBRA claims costs per enrollee were 63% greater than the claims costs of persons who remained employed (Exhibit 1, lower left arrow) – i.e. every COBRA enrollee had about 1.6 times the claims cost effect of a person that ultimately lost their job. However, there are (potentially many) more persons with good reason to fear job loss as there are (or will be) COBRA enrollees; accordingly, the effect of fear-of-job loss on MLR’s is (potentially much) greater than the COBRA effect during periods of significant job loss.
We weighted the per-person fear-of-job-loss and COBRA effects by the relative numbers of persons in either category, and in so doing estimate that during the most substantial periods of ’01 – ’03 job loss, that fear-of-job loss created twice as much pressure on MLR’s as COBRA (Exhibit 2). Given the relatively greater extent of job loss in the current downturn (Exhibit 3), we believe that job loss fears are at least twice as important as COBRA; and, that because such a larger proportion of the workforce has lost jobs, that the total degree of fear-of-job-loss MLR pressure is much greater now than in ’01 – ’03 (where we very roughly estimate an effect of 3.2%).
Critically, note that we do not need for employment to recover for the fear-of-job-loss effect to go away – we just need to stop shedding jobs. In other words, once gross job losses return to typical levels, we would expect the fear-of-job-loss pressure on MLR’s to abate. Clearly, going into 2010 fears of job loss should be much lower than in 2008 or 2009. Even if we were to take an aggressive view of peak unemployment in 2010 – say 11% – we’d emphasize that 5 of the 7 points of job loss between the ’08 employment peak and the ’10 trough have already been travelled (Exhibit 3, again).
Pulling the various threads together, we believe: that fear-of-job-loss has produced far more MLR pressure on the large capitalization insurers than COBRA enrollment; that this effect has largely played out; and, that by extension, insurers should see an alleviation of ’08 / ’09 MLR pressures as ’10 unfolds.
Why should a Public Option – if it’s not-for-profit and has to negotiate prices with providers – have a competitive effect that is any different from not-for-profit Blues plans? We examine the impact of not-for-profit Blues on state-level competition.
Both the House and Senate versions of reform bills contain public options; importantly, both versions require the public option to negotiate prices with providers – i.e., the public option, if it exists, almost certainly will not have access to Medicare payment rates, which was by far the biggest threat (to commercial insurers) associated with the public option. Moreover, the public option will have to charge rates adequate to cover its claims costs, cover its operating expenses, establish adequate reserves, and pay start-up capital back to the Treasury.
From the perspective of a commercial insurer, how is this any different than competing with a local Blues plan, especially if that plan is operated on a non-profit or mutual basis? We recognize the longer-term risk that, once a public option exists, the rules will change; however we believe that changing the rules enough to give the public option meaningful economic advantages relative to commercial insurers is a very large legislative task, much on par with passing a public option in the first place. In other words, we freely admit that the existence of a ‘Blues-like’ public option gives the government camel a nose in the tent, but we see the added risk as minor. Government is already more than 47 percent of total US health purchasing; and, from the commercial insurers’ perspective, there aren’t that many employees in this particular tent.
Over time, some Blues plans have converted to for-profit status, while others have remained as non-profits or mutuals. This sets the table for an interesting natural experiment – do non-profit Blues meaningfully disrupt competition in their markets, as compared to markets in which Blues compete on a for-profit basis? Using AMA data on health insurance market concentration and Consumer Union data on Blues status by state, we compared market concentration in states where the leading Blues plan is for-profit, with market concentration in states where the leading Blues plan is non-profit. The basis for comparison is the Herfindahl-Hirschman index (HHI) of concentration. Higher HHI numbers mean greater market concentration, which in turn implies less competition.
The 29 states with non-profit Blues have an average HHI of 3779, v. the 15 states with for-profit Blues which have an average HHI of 3319 (Exhibit 4; full state level detail in Appendix I). Thus the markets in which Blues are non-profit are less competitive (there’s more market share in fewer players’ hands), suggesting the Blues non-profit status is a meaningful competitive advantage (in all but one of the non-profit states, the Blues plan is the market leader). That said, the advantage isn’t terribly dramatic – the average market share of Blues plans in non-profit states is 54%, versus 50% in for-profit states.
This suggests that a not-for profit public option has some enrollment advantages over for-profit commercial insurers, but that the effect is modest. By no means does non-profit status appear to give an underwriter the power to wholly dis-intermediate for-profit underwriters.
We should further bear in mind that the public option is limited to participants on each state’s health insurance exchanges (HIE’s), and that participation on the exchanges is limited in whole or in part by considerations such as income, employer size (in terms of #’s of employees), and presence or absence of an affordable ESI offer from one’s employer. Thus the public option is somewhat limited to a sub-segment of the total market; and, as we showed in our previous call (Oct. 19, “Despite Near-Term Uncertainties, Health Insurers are Nearly 40 Percent Undervalued”), we believe that no more than 13 percent of the current ESI market would be eligible for and consider participation in an HIE, and of course further believe that not everyone would choose the public option if they did participate in an HIE.
Finally on the matter of the public option’s competitive impact, we note that CBO’s preliminary score of the House bill reflects CBO’s belief that the public option’s premiums would actually be higher than commercial premiums, for reasons of relative adverse selection (CBO assumes public option enrollees will be sicker). We have not seen the basis for CBO’s assertion, so while we can comfortably acknowledge that adverse selection in the public option is a feasible scenario, we can’t yet take a position on whether or not we believe this will in fact be the outcome.
We believe there’s more to 2010 (and beyond) pharmaceuticals pricing than meets the eye; PBM’s are the primary beneficiaries, followed by wholesalers.
The large-capitalization pharmaceutical manufacturers face considerable margin pressures in 2010 and beyond. Whether or not reform legislation passes, many of the group’s larger firms face patent expiry on key brands, and we note that with the exception of the Zocor expiry at MRK, both managements and the Street have a history of under-estimating the gross margin effects of generic competition.
If legislation does pass, the legislation’s provisions place considerable incremental margin pressures on pharma; these begin immediately in 2010, and extend well into the next decade.
Mandatory Medicaid rebates would increase from 15.1% to 23.1%. Medicaid is roughly 9 percent of US retail sales; accordingly the direct gross profit effect of additional rebates is roughly 1% on US sales. On top of this, manufacturers’ share of the brand industry’s tax represents roughly an additional 1% of US sales. Both effects begin immediately in 1Q/’10, assuming reforms pass.
We believe that the change in Medicaid rebates will serve to increase commercial rebates as well, and that this is a far more potent source of gross margin pressure than the Medicaid rebate dollars paid directly to the states. When the Medicaid mandatory rebate program became law in 1990, its effects were eased in over a couple of years; by 1993 manufacturers were required to give Medicaid 15.2 percent off of their average manufacturer’s price (AMP, roughly equal to the price at which manufacturers sell to wholesalers); or, their current best discount to commercial buyers, whichever was lower. In 1990, prior to the law coming into full effect, manufacturers’ average best discount was 35.8%; by 1995 manufacturers’ average best discount had fallen to 16% (Exhibit 5, see diamond-marked line at far left).
In 1995, Medicaid purchased 16 percent of the US retail prescription market; in comparison, the largest single commercial buyer had a purchase share in the (very) low single digits. Assume for the moment that your best commercial buyer has a purchase share of 2 percent; if you continue to agree to provide your 1990 discount level of 35.8 percent, you’ll have to give Medicaid the same price. Since Medicaid was 16 percent of sales, the added 20.6 percent discount above and beyond the statutory 15.2 percent minimum requirement would cost you 3.3 percent of your total US sales – i.e. saying yes to your 2% of sales best customer costs you 3.3% of sales. Better to say no.
Because of Medicaid’s most-favored-nation purchasing status, and its large purchases relative to any other buyer, manufacturers had every incentive to reduce commercial rebates to the Medicaid level of roughly 15 percent. This is exactly what happened; from Congressional Budget Office (CBO) testimony over the years, we know that manufacturers’ average best discount stayed around 15 to 16 percent from 1995 until at least 2003 (Exhibit 5 again, diamond-marked line). Not all manufacturers stopped at the 15 percent threshold, some brands in highly competitive categories were forced to rebate levels below the Medicaid threshold, and so wound up paying Medicaid rebates of greater than 15 percent. All in, the sales weighted average Medicaid rebate stayed at roughly 20 percent from 1995 to at least 2003.
In 2006, Medicaid’s purchase share of the US retail prescription market fell from nearly 19% to 9%, as Medicare Part-D pulled dual-eligible seniors out of the Medicaid drug program. Also from CBO testimony, we know that after years of holding steady at roughly 20 percent, the average Medicaid basic rebate had grown to 23.1% in 2007 – meaning that brands with more total sales dollars had exceeded the 15 percent Medicaid rebate threshold, that discounts on brands had gone further beyond the 15 percent threshold, or some combination of the two (Exhibit 5, square-marked line).
It’s easy to see how the disciplinary effect of the Medicaid rebate threshold would diminish as Medicaid’s purchase share fell by half. Despite this, our channel checks tell us that a substantial number of brands still stop short of offering rebates below the current 15.1 percent Medicaid threshold. Having run a pharmaceuticals pricing operation under the Medicaid best price rules for a number of years, we believe that manufacturers generally are able to negotiate price to commercial buyers with a feeling of confidence that the competition almost certainly will not make a rebate offer below the Medicaid threshold. In 2010, if reform legislation passes, this all changes – manufacturers will have reasonable confidence that competitors will not offer rebates below the new 23.1 percent threshold, but this marks a rather dramatic 8 percent drop in net pricing.
Because the Medicaid best price effect has so clearly served as an effective floor underneath pharmaceutical discounts, we see movement of the best price level from 15.1 percent to 23.1 percent ushering in a period of correspondingly steeper rebate levels. Returning to the top of our argument; this, plus the direct dollar cost of Medicaid rebates paid to states and the manufacturers’ tax (together about 2 percent of sales), creates substantial motive for manufacturers to accelerate top-line pricing as early as 1Q/10.
There is yet one more considerable piece of pricing pressure in the longer term. We’ve previously argued that the proposed (Senate version) Medicare Commission cannot meet its mandate of controlling Medicare cost growth simply by modifying payment rates to providers; rather, we feel the Commission almost certainly would have to come after innovators’ (such as brand drugs’) margins in order to come anywhere near fulfilling its mandate. We noted previously that nothing in the Senate bill’s language precludes the Commission from recommending reductions in input prices, thus there’s no reason to believe that the Commission would feel obligated to restrict itself to modifying providers’ payment rates, which is largely the limit of CMS’ current pricing authority. More recently, we noted that the bill’s language actually precludes the Commission from recommending lower provider prices during its first 5 years of operation (2015 to 2019); accordingly during this period the Commission would have virtually no other alternative than to reduce input prices. To our minds this is an incredibly potent source of future pricing pressure, and while we do not sense that Commission-related pricing threats are acknowledged by the capital markets, we are somewhat confident that manufacturers see the risk – all of which adds further to the argument that manufacturers have substantial incentive to accelerate real pricing.
All of the preceding speak to motive, which brings us to opportunity. We’ve made the comparison between now and 1998 (Exhibit 6) wherein manufacturers appear to have taken advantage of the considerable political distractions in this period. Whether reforms pass or not, the degree of political capital and legislative bandwidth consumed by the effort are simply staggering, despite the fact that health reform remains well below economic concerns on voters’ radar screens, and largely on par with Iraq / Afghanistan. From this, we infer that, after the health reform effort reaches an end, the near- to mid-term odds of Congress again turning its attention to anything related to healthcare, even for the express purpose of disciplining real price acceleration by pharmaceutical manufacturers, is as close to zero as it has ever been.
All in, this means we see high odds of both accelerated real (list) pricing by pharmaceutical manufacturers, and steepening commercial discounts. PBM’s are the obvious beneficiaries, being positively levered to both real price growth and deepening discounts. PBM’s contracts with pharmaceutical manufacturers typically include a low single-digit percentage ‘service’ fee on sales; this fee of course rises in real terms as real drug pricing rises, without any incremental cost required of the PBM. PBM’s contracts with plan sponsors and dispensing pharmacies typically embody some degree of spread on pharmaceutical pricing, i.e. the payments by plan sponsors to PBM’s are a slightly greater percentage of a pharmaceuticals’ cost than are the PBM’s payments to dispensing pharmacies for that same prescription. Again, in that it is reasonable to assume this is a stable percentage, the spread grows in real terms as real drug pricing rises.
PBM’s contracts with pharmaceutical manufacturers obviously include various rebates, and PBM’s contracts with plan sponsors generally require that these be shared in whole or in part. On average, across their entire book of business, PBM’s still capture a significant percentage of these rebates – thus as these rebates become larger percentages of list prices (which, again, are growing in real terms), the PBM’s stand to see real growth in the value of their retained rebates, again without any associated gains in operating costs.
As should already be very well appreciated, we also note that PBM’s are positively levered to patent expirations. Using data gathered by the Federal Trade Commission in 2002 and 2003 under subpoena, we can estimate dollar spreads on single-source brands and generics, according to whether the prescription was dispensed under the PBM’s payment authority in a retail outlet not owned by the PBM, or via mail from a mail-order pharmacy owned by the PBM. Using the PBM’s disclosures to weight the proportion of their PBM services business coming from either the retail or mail setting, we can calculate the average dollar spread for a single-source brand or generic across the PBM. In short, the larger mail is as a percent of total services, the more positively levered the PBM is to the transition from brands to generics (Exhibit 7). We recognize that only a minority of CVS’ profits are from PBM operations, but note that the firm’s retail pharmacy operations are also positively levered to the shift from brands to generics. The same FTC study found that at retail, dollar spreads on generics were more than 50% greater than dollars spreads on single-source brands.
Despite positive leverage to generics, and some odds of both accelerated real pricing and deepening commercial discounts, we note that consensus estimates of PBM gross margins are roughly static going forward (Exhibit 8).
Drug wholesalers are an obvious beneficiary of accelerated real pricing. Inventory management agreements limit exposure to real price gains, though we note that most agreements still allow wholesalers some fixed level of real pricing leverage above and beyond inventory on hand at the time of price increases. We note that PBM’s also have exposure to real price gains through their mail-order inventories. Unlike PBM’s wholesalers have no means of gaining from deepening commercial discounts.
- Consolidated Omnibus Budget Reconciliation Act (COBRA) of 1986 amended the Employee Retirement Income Security Act, and established employees’ rights to continued health coverage from their former employer for premiums not in excess of 102 percent of employed persons’ premiums in their former firm.
- This remains true even after accounting for the impact of subsidies given to laid-off workers for whom premium costs exceed certain thresholds of income, which were not available in previous down cycles. It’s quite clear these subsidies should increase the number of COBRA enrollees and thus COBRA-related medical loss ratio (MLR) pressures, though it’s important to recognize that as subsidies become more generous, the marginal COBRA enrollee almost certainly becomes healthier.↑
- Note the difference in consumption between COBRA enrollees who remain enrolled for the full year, v. those that lose COBRA coverage (and do not return to ESI) in a given year. We expect that COBRA enrollees having no employment (and thus no ESI) prospect accelerate their consumption as the end of their coverage approaches. We recognize that the rate of excess claims for COBRA v. employed does not follow the same pattern as it does for persons that lost their jobs v. those that remained employed – i.e. greater in the downturn, and lower in the upturn. We emphasize that economic cycle should have no effect on per-person COBRA claims costs, other than increasing the odds that a greater number of persons reach the end of their COBRA coverage without an ESI to go to, and so accelerate their consumption – as happened consistently across both the down and up cycles.
- Those who pay premiums effectively own the plan.
- The Herfindahl-Hirschman index (HHI) is the sum of the squares of market shares for the top 50 participants in a market; or, where fewer than 50 participants exist, the sum of the squares of market shares for all participants. The index in this example assumes that market shares are expressed as whole numbers rather than as percentages (e.g. 50, rather than 50%).
- The reason the sales weighted best price discount (lowest price to comm’l buyers) reflects lighter discounts than the basic rebate (paid to Medicaid): not all brands provide significant discounts to commercial buyers, but all must pay at least 15 percent to Medicaid – so a brand with a 5 percent commercial discount would tend to move the best price discount average to a lighter discount level, even though the brand pays 15 percent to Medicaid. Conversely, a brand with a discount beyond the 15 percent threshold would raise both the average best price discount and the average basic rebate. Separately, note that total rebates to Medicaid are much higher than the basic rebate, as total rebates also include a true-up for real price growth. We ignore this for the present argument, as it has no bearing on either pricing or discounting behavior.
- N.B.: the Commission is not allowed to reduce eligibility, reduce benefits, or raise premium costs.