Three Reform Realities That Aren’t Priced In

Print Friendly, PDF & Email

Richard Evans


March 29, 2010

Three Reform Realities That Aren’t Priced In

  • Large capitalization health insurers trade at a 25 to 30 percent discount to the SP500; this valuation implies some level of dis-enrollment and margin erosion, where we see rising enrollment and stable to expanding margins. Closer in, we continue to believe claims cost expectations are too high for insurers in 2010 / 2011. Health insurers remain our preferred healthcare sub-sector by some margin
  • We expect PBM gross margins to expand in the near term, as commercial discounts from brand drug manufacturers grow larger as an indirect consequence of reforms. We retain our conviction that PBMs are subject to gross margin erosion in the mid- to longer-term as patent expiries accumulate on ‘traditional’ retail brands
  • The Independent Medicare Advisory Board is perhaps the most consequential element of reform legislation, and the least appreciated. Beginning in four years, the Board is required to hold per beneficiary Medicare cost growth within an ever-tightening range of GDP or CPI – without reducing eligibility or benefit levels, without adding to beneficiaries’ premiums or cost sharing, and for the first half decade, without reducing payments to providers. These restrictions so narrow the Board’s latitude that it has no option but to attempt to lower input costs, particularly those associated with innovative products. Accordingly, we expect innovators’ margins will begin to come under considerable pressure on sales to Medicare

Summary and Conclusions

With large-scale federal health reforms having passed into law – despite our expectation to the contrary – we see large capitalization insurers as (still) undervalued, believe that PBMs are likely to see a near-term gain in gross margin before gross margin pressures emerge in the mid- to longer-term; and, believe that the mid-term effect of reform provisions on innovators’ margins are under-appreciated. For insurers, in the near-term we believe claims costs will be lower than expected, and in the longer-term believe that reforms offer more positives and fewer negatives than suggested by valuations. For PBMs, notwithstanding our concerns of gross margin pressure in the mid- to longer-term, we expect gross margin gains in the short term as commercial rebates on pharmaceutical brands grow larger, as a consequence of steeper Medicaid discounts called for in the reform package. Separately, we re-emphasize our concerns with respect to the Independent Medicare Advisory Board (formerly the ‘Medicare Commission’) and its likely effect on innovators’ margins. The Board is now law, and while we ultimately believe that it cannot meet its statutory mandate of keeping Medicare spending within a fixed range of growth in the broader economy, we certainly expect it will try, and expect innovators’ margins to come under fire (a soon as 2014) as a result.

Health Insurers and Reform

Changes in Enrollment

Reform-related changes in enrollment are the net of current employer-sponsored enrollees being lost to private purchase of health coverage on the exchanges, and gains in enrollment as a result of federal subsidies for purchasing health coverage.

Potential for erosion of the employer-sponsored insurance (ESI) market by health insurance exchanges (HIE’s)

The immediate (pre-2017) scope of the health insurance exchanges (HIE’s) allows for insurers of 100 employees or less to purchase their small group coverage on an HIE – and roughly 24 percent of employer-sponsored insurance (ESI) premiums are paid by businesses of fewer than 100 employees (Exhibit 1). Thus in theory if all sub-100 employee firms leave their current ESI providers and instead purchase coverage on an HIE, and if their current ESI providers do not participate in the HIE, 24 percent of ESI premiums would shift away from the insurers that underwrite the ESI business in today’s market. Clearly this overstates the likely case, as current ESI providers are almost certain to offer coverage on HIE’s, and so will capture some (probably very large) portion of the employers who switch to the HIE’s.

Separately, some portion of employees may choose to refuse the ESI offered by their employers, and instead either go without coverage (and pay the penalty), or purchase coverage on an HIE. Many would not be eligible for subsidies in the HIE – especially those whose employers offered affordable ESI; and, of those that are eligible for subsidies, not all have an economic incentive to choose the HIE over ESI. For reference, the reform package offers subsidies up to 400% of the Federal Poverty Level (FPL), which is about mid-way through the nation’s third income quartile. We estimate that roughly one-third of ESI premiums are paid by or on behalf of employees (and employees’ households) having incomes up to 400% FPL, or the mid-point of the third quartile (Exhibit 2). For households in these income ranges, the economic incentives can break either way (keep your ESI; or, refuse ESI and buy coverage on an HIE). We estimated the economic effect of refusing ESI and choosing HIE – assuming the employee was subsidy-eligible – by household income. We looked at economic effect two ways – ‘total’, and ‘cash.’ Total economic effect takes into account the cash gains to the employee (they no longer have to pay their share of ESI premiums) net of both the HIE premium after subsidy, and the lower actuarial value of HIE coverage relative to ESI coverage. Cash effect looks only at the added cash flow the household enjoys as a result of no longer paying its share of the ESI premium, net of the payments made to participate in the HIE (Exhibit 3). Households up to $50,000 annual income (2013 dollars) gain cash flow by choosing the HIE, but only households up to about $30,000 annual income see a total economic gain from choosing the HIE over ESI. A dollar of cash is generally more attractive to lower-income households than a

dollar of benefit – irrespective of the relative economic values – accordingly we fully expect some portion of households to sacrifice total economic value for added cash flow. Our rough conclusion is that households with incomes at or below $40,000 are relatively more likely to prefer HIE’s over ESI, and the opposite is true for households with incomes above $40,000. This income level is about half-way through the nation’s second income quartile (Exhibit 2, again); we estimate roughly 13% of ESI premiums are paid by or on behalf of these households.

Thus insurers’ dis-enrollment risks have more to do with employees refusing ESI than with employers moving to the exchanges. If employers leave their current ESI provider and move to an HIE, they move to a small group HIE – which we expect to be a highly functional underwriting market where the large capitalization insurers write profitable coverage. In fact – as a small employer – we expect the efficiencies created by the small group HIE’s to bring more employers in, raising demand for ESI.

In contrast, we believe that the HIE’s for individual coverage ultimately may fail because of adverse selection, unless Congress either reforms the individual market (e.g. by strengthening individual mandates), raises additional revenues to pay for spiraling premiums, or institutes outright price controls. Large capitalization insurers still are likely to be in these markets for whatever period they are functional, but because we believe the individual coverage HIE’s ultimately may fail, we’re somewhat concerned that a portion of employees leaving ESI for HIE’s may be lost to the large cap insurers. Thus to our minds the worst-case scenario is for 13 percent of ESI premiums to ‘leave’ ESI for an ultimately dysfunctional individual coverage HIE. And this is truly worst case, as the 13 percent figure assumes that everyone below the $40,000 income level is lost to the large cap insurers, and ignores the increase in ESI demand from smaller employers that become first-time sponsors of coverage in the small group HIE’s.

Enrollment gains are significant, but are tempered by economics, attitudes, and adverse selection

Those households that cannot purchase coverage as part of a small group have the option of purchasing coverage on the individual-coverage HIE’s. We’ve estimated the net gains in private insurance coverage as the HIE’s begin operating, and as households that were previously un- or under-insured become eligible for subsidies. Setting aside for the moment our expectation that the individual-coverage HIE’s ultimately cannot survive adverse selection, the rate-limit on enrollment in the individual-coverage HIE’s has much to do with the economics and attitudes of the households involved.

As regards economics, for a very large portion of the relevant population, the difference between paying premiums and paying the penalty for being un-insured is equivalent or nearly so to the payments on a respectable new car (Exhibit 4). And as regards attitudes, consumers in the relevant income strata (133% – 400% Federal Poverty Level (FPL)) place relatively low values on health insurance – which suggests that at comparable prices, they’d prefer the new car. The recently published 2007 Medical Expenditure Panel Survey (MEPS) update makes available a substantial set of responses (18,000 non-institutionalized US adults age 18 and older) to questions that get to the heart of consumer attitudes regarding health coverage, or even health care. To begin with, insured and uninsured respondents’ perceived health status is the same, which eliminates an obvious source of potential response bias (Exhibit 5). As compared to those with insurance, the uninsured are considerably more likely to feel that they either do not need health insurance, do not need health care, or that health insurance is not worth the cost (Exhibit 6). Comparing these exhibits, it’s fairly clear that a substantial proportion of the uninsured are uninsured by choice; which, in light of the considerable expense of purchasing insurance further suggests that large numbers of the uninsured will choose not to purchase subsidized coverage, and will instead pay the penalty[1]. To gauge how this might play out, we crossed responses to each of the four questions with ‘un-insurance’ rates for negative responses; this gives us a series of rough estimates for the numbers of persons that might choose to remain uninsured, even if reforms pass (Exhibit 7).

Taking the lowest and highest results, we very roughly estimate that 21% to 52% of those presently uninsured will choose to remain uninsured even after reforms are put in place. Adjusting this for uninsured respondents in the most relevant income ranges (133% – 400% FPL); we estimate that 27% to 41% of eligible enrollees would choose to remain uninsured. At a minimum this strongly suggests that reforms will leave substantial numbers of persons without coverage. Worse, for those that choose to purchase individual coverage on the exchanges, ultimately they face high odds of escalating premiums due to adverse selection – sicker persons are more likely to enroll than healthier persons, driving premiums higher and subsequent enrollment lower in a self-fueling spiral. Underwriters can no longer deny coverage for pre-existing conditions, thus with the exception of having to wait for the annual enrollment period, the door is open for households and individuals to enter and exit the market as they become ill, or get well. In the final legislation, premium subsidies are indexed to keep pace with the excess of premium cost growth over wage growth – but only so long as the total cost of premium subsidies remains within the revenue-generating provisions of the law. As presently designed, we’re convinced health costs will run well ahead of wages, and that Congress will have to reduce premium subsidies relative to total premium costs, raise additional revenues, control prices, or some combination of these. Assuming for the moment that premium subsidies keep pace with premium growth, we would expect subsidies to reduce the number of uninsured from roughly 16 percent of the population pre-reform, to between 9 and 12 percent of the population post-reform, the result being enrollment gains of between 14M and 21M in 2014.

Changes in Underwriting Risk

The Effect of Underwriting Restrictions

Much is made of the restrictions placed on insurers by federal reforms – for example those having to do with age-related premium differences, rescission, lifetime coverage limits, and pre-existing conditions[2]. In spite of this, we continue to believe that these restrictions change the nature of what is being sold, but not the nature of competition among sellers. In the pre-reform market, any single insurer with more permissive underwriting standards would immediately be subject to adverse selection, as riskier enrollees moved to that insurer for lack of alternatives. By forcing all insurers toward more permissive underwriting standards – that are uniform across insurers – the potential for relative adverse selection among insurers is all but eliminated.

But the magnitude of absolute adverse selection, i.e. the average risk of an enrollee, certainly will increase. As a result of higher average per-enrollee risks, premiums should rise, assuming competitive intensity remains constant – which it should, as the reforms do nothing to alter the structure or intensity of competition among underwriters. Accordingly, we expect that underwriting restrictions will serve to reduce medical loss ratios (MLRs) at least slightly, as underwriters seek to offset greater risk with higher margins.

We reiterate our view that the present insurance model is a cost plus business – for the time being insurers compete on mark-ups to the underlying cost trend, rather than competing on the basis of lowering the underlying trend. By changing the nature (costs and risks) of what is sold, pending insurance reforms inevitably raise premiums[3]. By not changing the structural determinants of competitive intensity, pending reforms do little if anything to reduce risk-adjusted MLRs.

MLR Restrictions

Federal reforms provide for medical loss ratio (MLR) minimums on private insurers; these appear to apply to all medical coverage sold in the US, whether sold through the exchanges or otherwise. The minimum ratio for large groups is 85; for individuals and small groups the minimum is 80. These restrictions go into effect in 2011 and do not sunset.

Importantly, the bill calculates MLR differently than insurers’ calculate and report MLR in financial disclosures. Specifically, the revised Senate bill deducts federal and state taxes, additions to reserves, regulatory fees and other charges from revenues, thus reducing the denominator (and raising the ratio) relative to the traditional reported calculation. And, the revised bill adds to the claims cost numerator (again raising the ratio) that portion of operating expenses incurred for ‘activities that improve healthcare quality’. Such activities are not defined in the bill; the National Association of Insurance Commissioners (NAIC) is tasked with developing this definition by 2011.

We examined the simple-weighted average / as-reported medical loss ratios for the four largest US insurers from 2006 through 2008, and compared reported MLRs to those that we would expect to see under the calculations envisioned by the Senate bill (Exhibit 8). As-reported MLRs varied from 80 to 82 percent between 2006 and 2008. Adjusting revenues for taxes (2.4 percent of revenues) and additions to reserves (1.1 percent of revenues) raises the MLRs by roughly 3 percent. The portion of past operating expense that would fit NAIC’s as-yet non-existent definition of course cannot be known, so we consider a range. Note that we only have to assume that one-quarter of operating expenses can be attributed to activities that improve health quality in order to show MLRs that are already consistent with the most restrictive (large group, 85 percent) of the Senate’s proposed MLR ranges. The ‘blended’ MLR requirement will of course lie somewhere between 80 and 85; and, we make no assumption for expanded volumes as a result of federal subsidies. We note that early evidence supporting our hypothesis emerged last week when WellPoint issued MLR guidance for 2010 which is 170 bp higher than 2009. Importantly, the majority of this increase was attributed to reclassification of some benefits (e.g., nurse hotlines and disease management) which would presumably count as ‘quality improving’ under terms of the legislation.

Effect of Tax Provisions

‘Cadillac Tax’ Effectively Eliminated

Recall that the Senate-passed ‘un-reconciled’ bill imposed an excise tax of 40 percent on so-called Cadillac plans (individual premiums of $8,500 per year; family premiums of $23,000; growing annually) beginning in 2013. We had argued that the steep excise tax effectively formed a brick wall through which premiums would not pass. The impact of the provision was clearly a deceleration of revenue growth for insurers but not a revenue decline over the foreseeable future. However, our belief is that such a deceleration has been inevitable for some time for simple macro-economic reasons – as health costs grow faster than wages, the cash left over after paying insurance premiums has declining purchasing power. For an increasing number of households the health cost / wage tradeoff is no longer considered worthwhile; over the next decade, as health cost growth continues to exceed wage growth, this ultimately becomes true for the population in general. Thus in our view, the excise tax would simply have codified a specific rate of decline in health premium growth relative to wage that would have happened anyway.

The reconciliation package rather dramatically minimizes the impact of the Cadillac tax by both delaying its initiation (from 2013 to 2018) and raising the thresholds ($10,500 for an individual from approx. $9,410; $27,500 for a family from approx. $26,483[4]). Organized labor, the most vocal opponent of the tax, supported the change and their announced intention to continue to fight it; the 8 years they have to wage that fight, and the loose legislative language underlying the tax cast doubt about whether it will ever be implemented. Thus the reconciliation language reinforces our thesis that the Cadillac tax is a non-issue for insurers – macroeconomic pressures will limit premium growth relative to wage long before the tax can have an effect.

Excise Tax on Insurers Delayed Further

The original draft Senate bill (“The Baucus Bill” circulated in September 2009) provided for a $6.7B tax on insurers beginning in 2010; the enacted bill postponed the beginning of such taxes to 2011, and employs a rising scale; the reconciliation bill postpones the beginning even further, to 2014, though employs a significantly steeper rising scale. Certain non-profits are exempt from paying the tax; it is not yet clear to us whether non-profits’ premiums are included in the denominator used for determining for-profits’ share of total fees. If not, this means that the average for-profit insurers’ share of the tax could be much larger than in the original bill. We continue to view the tax as a pass-through, but recognize that the change toward having a tax that is paid only by for-profits puts non-profits at an incremental underwriting advantage as the tax escalates. Even assuming for-profits pay the entire $14.3B, at the first year of ‘full’ tax (2018), and assuming a premium growth CAGR of 8 percent, the tax would represent a less-than 2 percent cost of underwriting dis-advantage for for-profit insurers relative to not-for-profit insurers. The effect is even smaller in years preceding 2018 because of the ramp-up in the tax, and in subsequent years as the tax remains nominally fixed.

What’s Implied at Current Valuations

We favor insurers for two fundamental reasons: Near-term, we believe the market is extrapolating ‘08/’09 claims costs into ’10, and in so doing is missing that while fear-of-job-loss added a great deal to ‘08/’09 claims cost, these pressures are absent in ’10 – thus our immediate expectation of improving gross margins[5]. Further out, we believe the market prices the enactment of reforms as a net negative, where we see a net positive – enrollment increases; and, with no change in competitive intensity, added costs are effectively passed-through.

We use the SP500 as a long-term valuation benchmark. We assume 5.6% (annual) long-term SP500 EPS growth; this rate is the product of 2.2% GDP growth, 2% CPI growth, and 1% productivity growth. We discount long-term earnings projections at 9.6%; this figure is equivalent to long-term (1926-2009) nominal returns on US equities of 10.57% at average inflation of 2.99%, adjusted for CBO’s long-term inflation forecast of 2%.

Against this backdrop, we make a series of assumptions regarding rates of premium growth and changes in insurers’ cost structures, and compare the relative values of the resulting earnings streams. In all scenarios we assume that premium growth tracks medical inflation until we approach the point at which households’ non-health wage growth approaches zero, after which premium growth tracks with wage growth. We then make any combination of three binary assumptions, to produce a total of 8 scenarios: 1) either no or 13% dis-enrollment from ESI as HIE’s emerge; 2) either no change or a 20% drop in insurers’ net margin; and 3) we assume that either an additional round of reforms ‘resets’ HC spending as a % of GDP to year 2000 levels, and that these reforms unfold from 2020 to 2025 – or that no further reforms take place. Please note that our expectation regarding enrollment is for enrollment gains, though in the interest of a more conservative valuation we model only constant (pop’n adj’d) or declining enrollment.

The results are in Exhibit 9, and scenarios with relative values (insurers v. SP500) in line with current valuations are highlighted. Health insurers trade at a 25 – 30% discount to the SP500 on 2011 / 2012 PE’s; current valuations appear to anticipate some combination of dis-enrollment, margin pressure, or an eventual and severe (the scenario we model cuts per-capita health spending by fully one-third) round of further reforms. We view these scenarios as too severe, thus our conclusion that health insurers remain undervalued.

PBMs – Gross Margins May Gain Short-Term; Mid- to Long-term Gross Margin Pressures Remain

We recently published our thesis that PBMs appear to be at or near the peak of their pricing cycle[6]. In short, we argued that PBMs largely function as buying clubs that allow plan sponsors to access steep rebates on traditional chronic-use drugs; and, that this business model relies on a market wherein a large proportion of sales come from traditional brands that compete within a category of largely interchangeable alternatives (e.g. statins). We showed that a very large proportion of rebates comes from a very small number of traditional products, and that 75% of the dollar sales of these products will have been lost to generics by 2013. Recognizing this, PBMs are transitioning from traditional to specialty services; we argue that profits from specialty sales and services will not be large enough or come fast enough to offset falling traditional sales. Finally, we argue that the management of traditional drug benefits – very large numbers of very brief (a patient at the retail drug counter) pharmacist-focused encounters with relatively low dollar value and clinical complexity – is quite different from the management of specialty drug benefits – relatively low numbers of slower moving physician-focused encounters with very high dollar value and substantial clinical complexity; and, that PBMs’ HMO clients are nearly as well-suited to manage the specialty drug benefit.

All this aside, we also noted that PBMs benefit when rebates on commercial drugs grow larger, since PBMs are able to keep some portion of manufacturer rebates. We’ve showed in the original PBM call[7] that drug rebate levels are heavily influenced by Medicaid best price dynamics – Medicaid is the single largest retail drug purchaser, and gets the lower of the lowest price in the market or a 15.1% discount. As a result, manufacturers hesitate to offer rebates below the 15.1% best price threshold, since this creates a very large incremental rebate obligation to Medicaid. In 2006, Medicaid’s share of retail purchases fell by nearly half as dually eligible[8] seniors went to Part-D. This had the effect of tempering the best-price consequences of deeper commercial rebates, and as expected, commercial rebates grew larger (Exhibit 10). PBMs gross-margins benefitted; 2006 is the only year in recent history when PBM gross-margin growth de-coupled from the usual drivers – real drug price growth, and generic mix (Exhibit 11). The reform measure immediately lowers the Medicaid mandatory rebate from 15.1% to 23.1%; accordingly we expect brand manufacturers’ commercial rebates again will grow larger – to the benefit of PBM gross margin.

Thus all in, despite our conviction that PBM gross margins are likely to decline in the mid- to longer-term, we expect a one-time, near-term benefit to PBMs’ gross margins.

The Independent Medicare Advisory Board

As we’ve highlighted before, the reform package includes an Independent Medicare Advisory Board whose task, beginning in 2014, is to make proposals that constrain the per-enrollee growth in Medicare spending to an increasingly tight range relative to indices (CPI, GDP) of growth in the broader economy. The law requires Congress to consider the Board’s recommendations under expedited parliamentary procedures; the net effect is that Congress is pressured to either accept the Board’s recommendations, or come up with its own means of achieving the same or similar degree of cost savings.

In the current market, the Center for Medicare and Medicaid Services (CMS) sets the price of Medicare and Medicaid services almost exclusively by setting payment rates to hospitals and physicians. Only with very rare exception does CMS have direct or near-direct influence on the price of inputs to care (e.g. Epogen). We emphasize that the Independent Medicare Advisory Board is under no such constraint – in fact quite the opposite – from 2014 to 2020 the Board is largely enjoined from recommending changes to providers’ payment rates. The Board is further enjoined from restricting eligibility, reducing beneficiaries’ access to benefits, or requiring beneficiaries to pay more for their benefits.

Given the strength of forces that drive health cost growth faster than GDP, the Board’s economic mandate to reduce per-beneficiary cost growth arguably cannot be met – even if it had the (pre-2020) authority to reduce provider payments. Accordingly we see practically all of the Board’s cost-reduction recommendations – especially before 2020 – being focused on the cost of inputs to care. Reducing the cost of commodity inputs (e.g. prescription generics) is an unlikely path; arguably competition will have reduced unit prices to cost of capital plus a small margin; even if it hasn’t, the margins are too small to be meaningful in the context of the Board’s goals. That leaves only one thing – innovators’ margins – thus our expectation that innovators’ margins come under immediate and heavy pressure as the Independent Medicare Advisory Board begins operating in 2014.

  1. By no means are we arguing that being uninsured is only or even predominantly a matter of choice; we believe that an enormous proportion of the population both needs and wants health coverage. Nor are we arguing that preserving one’s ability to choose to be uninsured is without considerable system-wide socio-economic costs. For our broader arguments on the health care system and prospects for reform, please see “Health and Capital”,
  2. Bans of denial based on pre-existing conditions in all employer sponsored plans would begin in 2014 under the reconciliation bill.
  3. As presently designed, we would in fact expect the health insurance exchanges to encourage insurers to offer higher rather than lower priced coverage – potentially expanding MLRs. Presumably many of the exchange participants will be subsidized, meaning that at least in the first year, premiums are capped as a percentage of income. If the ‘list’ price of plans is in excess of my percent of income limit, e.g. 10 percent, all of these plans are priced the same to me – they all cost me 10 percent of income. In such a case, I am more likely to choose the $12,000 plan than the $11,000 plan, all else being equal – as I would tend to think I was getting a better deal. By no means are we arguing that this makes any socio-economic sense whatsoever – we’re simply arguing that, as designed reforms appear more likely to raise, than lower, insurers’ margins.
  4. 2013 thresholds were to rise annually at CPI+1% under the Senate bill.
  5. “Health Insurers 40 Pct Undervalued” Sector & Sovereign, LLC October 20, 2009
  6. “PBM Gross Margins – This Looks Like the End of the Cycle” Sector & Sovereign, LLC March 10, 2010
  7. Ibid #6
  8. i.e. eligible for both Medicare and Medicaid
Print Friendly, PDF & Email