Despite Near-Term Uncertainties, Health Insurers are Nearly 40 Percent Undervalued
October 20, 2009
Despite Near-Term Uncertainties, Health Insurers are Nearly 40 Percent Undervalued
- We modeled reform-related and macro-economic threats to health insurers’ earnings power over the longer-term, and compare the present values of insurers’ longer-term earnings under various scenarios to a baseline assumption of SP500 earnings power.
- All of our scenarios assume deceleration of health insurers’ top-lines to GDP rates of growth by 2020, and no revenue growth in excess of GDP at any point thereafter. Whether or not legislation passes, we believe this is correct, as any more generous assumption produces logarithmic declines in real non-health purchasing power beyond 2020.
- Only by further assuming that net margins fall to 20% below their long-term average, that 13% of the employer-sponsored market is lost to the commercial health insurers when reforms become active in 2013; and, that a subsequent (2020) reform effort rolls back health spending by nearly 10 percent of GDP, do we begin approaching earnings scenarios that justify current valuations.
- Not only do current valuations incorporate all of these (to our view, extreme) pressures, valuations assign zero probability to a legislative failure, zero probability of productivity gains from a decade of residual real pricing power, and zero probability of enrollment gains to insurers from government subsidies.
- We acknowledge the immediate pressures on sales and margins from both reduced enrollment and comparatively higher claims in 2009, and the uncertainty around margins attributable to these pressures and the possible initiation of the $6.7B tax on insurers in 2010. Despite these near-term pressures, we believe earnings power in the mid- to longer-term is undervalued by nearly 40 percent.
The health insurers’ business model faces a number of threats: longer term limits on premium growth as insurance premiums represent an ever greater portion of total wages; relatively weak enrollment and relatively high per-beneficiary claims cost related to changing employment in the near- to mid-term; and, specific earnings threats that tie to current health-reform efforts.
Relative performance and valuation has suffered as a result; on a cap-weighted basis insurers trade at a 58%, 50% and 48% discount to the S&P500 on 1, 2, and 3 year estimates (Exhibit 1).
Modeling of Reform-Related Threats to Insurers’ Earning Power
We modeled reform threats to insurers’ earnings using the Senate Finance committee’s bill as a proxy for health reform effects, though our valuation conclusions acknowledge and account for the likelihood of this bill shifting left, and for the likelihood of future health reforms.
Broadly speaking, the Senate Finance bill contains two threats to insurers’ longer-term earnings power, one on the ‘low’ end of the business, and one on the ‘high’ end. On the low end, insurers face some risk of losing current employer-sponsored insurance (ESI) beneficiaries to government-sponsored health insurance exchanges (HIE’s) with or without a public option. On the high end, premium growth is potentially at risk from the effects of excise taxes on Cadillac premiums. We modeled each of these effects.
Share of the Employer-Sponsored Insurance (ESI) Market by Firm Size
Recall that the Senate Finance bill calls for two distinct markets to be created in health insurance exchanges (HIE’s), one for individual coverage, and another for group coverage. Quoting from the amended bill (p. 20): “Beginning in 2015, states must allow small businesses up to 100 employees (to) purchase coverage through the SHOP (i.e. group) health insurance exchange and states may allow employers with more than 100 employees into the state exchanges beginning in 2017 (emphasis and parenthetical statement added).” Accordingly the immediate (pre – 2017) scope of the group HIE’s is limited to businesses of 100 employees or less; after 2017 this constraint is lessened to the extent states choose to allow larger businesses into the group HIE’s. By researching employment, wages, and health benefit offer and acceptance rates by firm size, and by making reasoned
assumptions regarding relative premium costs by firm size, we can estimate the percent of the employer-sponsored insurance (ESI) market attributable to firms of various sizes. The result is in Exhibit 2; we estimate that 24 percent of ESI premiums are paid by employers and employees in firms having 100 or fewer employees. Thus in theory, if all of these employees left ESI for coverage under an HIE; and, if none of these employees enrolled in a commercial insurer’s offering on an HIE (i.e. the insurers chose not to participate, or were wholly dis-intermediated by a public option), then commercial insurers would lose 24 percent of the ESI market.
Share of the ESI Market by Employee Income
Not all of these employees would be eligible to leave ESI for government subsidized coverage in an HIE, nor would all of them want to if they could. Again quoting the bill (p. 38): “As a general matter, if an employee is offered employer-provided health insurance coverage (ESI), the individual would be ineligible for a low income premium tax credit for health insurance purchased through a state exchange. An employee who is offered coverage that does not have an actuarial value of at least 65 percent or who is offered unaffordable coverage by their employer, however, can be eligible for the tax credit. Unaffordable is defined as 10 percent of the employee’s income (emphasis and parenthetical statement added).”
We assume that an employee having little or no interest in health insurance is presently un-insured. Accordingly the downside risk of HIE’s on ESI demand is the question of whether employees who have some interest in insurance choose to continue purchasing through their employer, or reject their employer’s insurance option in favor of purchasing a government subsidized option on an HIE.
HIE’s eventually provide subsidies from 100% to 400% of federal poverty level (FPL). To frame matters broadly, we model the percent of the ESI market attributable to income quartiles; 3 percent of ESI premiums are from the lowest quartile, 20 percent from the second, 33 percent from the third, and 44 percent from the highest income quartile (Exhibit 3). For reference, the 400% of poverty cut-off (for average sized households) is roughly half way through the third income quartile, i.e. in very rough terms about a third of ESI premiums come from households with incomes at or below 400% of FPL (i.e. all of the first and second quartiles, and roughly half of the third).
To better understand which of these households might abandon ESI for an HIE, we modeled the economics of the decision across the relevant income ranges. The decision has several moving parts – changes in the value of health benefits received, changes in after tax salary, and changes in the total economic value of the compensation package (we include government sponsored coverage in the total comp calculation for comparison purposes). Households leaving ESI for an HIE save the amount of their employee contribution to ESI premiums (and so may increase their cash compensation if their contribution to HIE premiums is lower), but receive less valuable health benefits, as HIE benefits are on average less generous than ESI benefits. The results are in Exhibit 4. Households with (2013) incomes less than roughly $32,000 increase the total economic value of their salary + health benefits package by choosing the HIE over ESI, largely because of the balance between high government subsidies for HIE participation and low employer subsidies for ESI participation at and below this income level. Note however that households having incomes up to roughly $50,000 increase their after-tax cash compensation by leaving ESI for an HIE, though the lesser value of HIE benefits relative to ESI benefits means the total economic value of their salary + health benefits package is lowered by rejecting ESI.
Exhibit 4’s detail is useful for reference; Exhibit 5 models households’ decision more clearly. Here we show the change in value of cash and total compensation if a household abandons ESI for an HIE. We also show the percent of income paid for health insurance coverage for an average plan by households in each income bracket. Recall that employees offered ESI for premium contributions below 10 percent of income cannot get subsidized coverage on an HIE; also keep in mind that since employers often offer a range of plans – some cheaper than others – that some proportion of households facing contributions for the average plan that exceed 10 percent of income still may have an available ESI option that is below this threshold.
To our minds, household incomes of roughly $40,000 are a critical breakpoint (the vertical line in Exhibit 5). Above this income level, the added after tax cash of rejecting an ESI becomes relatively small, the total economic loss from rejecting the ESI becomes relatively large, and the odds of having an ESI option with premiums below 10 percent of income becomes greater. Conversely, below this income level, households see relatively larger after tax cash gains, relatively smaller total economic losses from rejecting ESI (and in many cases gains), and are less likely to have an ESI option that costs them less than 10 percent of their income. We conclude that households with (2013) incomes at or below $40,000 are relatively likely to abandon ESI for government subsidized coverage in an HIE, and that households with incomes above $40,000 are relatively likely to stay with ESI, assuming reasonable coverage is offered. Glancing back to Exhibit 3, this $40,000 threshold is about half way through the second income quartile, which represents 20 percent of the ESI market on a top-line (i.e. premium) basis. Assuming every household below $40,000 leaves ESI, this reduces ESI premiums by 13 percent (half of the second quartile, and all of the first quartile).
Net Effect of Firm Size and Employee Income on Demand for ESI
This leads to the question of whether we should assume that all employers sized 100 employees or fewer leave ESI, and/or whether we should assume that all employees with household incomes of $40,000 or below leave ESI. In short, no to the former, and yes to the latter.
Firms of 100 or fewer employees can purchase coverage on an HIE – and many of them should – but this doesn’t mean that these firms are ‘leaving’ ESI. They’re simply buying their ESI on an HIE. Recall that the bill configures individual and group HIE’s as separate markets. Also recognize that our concerns regarding adverse selection – i.e. our belief that healthy consumers shopping for coverage in the individual markets will jump in and out of coverage as needed – are much less relevant in the distinct HIE for group coverage. State laws for group coverage already provide for many of the underwriting rules envisioned in the Senate Finance bill, such as guaranteed issue. Moreover, groups tend not to jump into and out of coverage based on need; rather, groups tend to be blocked from coverage by high benefit costs in the broad market, and/or by high risk ratings attributable to their own claims history. The Senate Finance bill makes it difficult for insurers to individually rate groups, meaning the group HIE functions as a larger risk pool. This is a more efficient underwriting environment, which means groups with high prior claims costs can get back in, boosting total demand. Insurers, free to price their broad offerings within the group HIE, can price in the risk of less healthy groups re-entering the market. Insurers’ per-group administrative costs may be lowered for HIE participants as well, as there is an opportunity to standardize administration across client firms. Finally, smaller employers with relatively lower wage employees are eligible for subsidies, which on net should increase demand for coverage on the group HIE. Thus we do expect smaller employers to purchase on the group HIE’s, but we expect commercial insurers to be there writing the policies, and we expect the business to be profitable. Accordingly we see no need to penalize commercial insurers’ earnings power for the prospect of smaller employers buying coverage on an HIE.
In contrast, we clearly do expect some number (13 percent or fewer) of employees to leave ESI for coverage in the individual (i.e. non-group) HIE. Whether or not commercial insurers are there to pick up the business is an open question.
We’re convinced these markets eventually fall apart under adverse selection pressures unless they’re re-designed; which, by the way, has the effect of making ESI (no adverse selection run-away) relatively more attractive than the individual HIE once adverse selection pressures unfold. In other words lower income households may leave ESI for an individual policy in an HIE when the HIE’s begin operation, but if our views on adverse selection play out, many are likely to return to an ESI. Despite the great risk of adverse selection commercial insurers have some incentive to offer individual coverage on the HIE’s as they begin operation – i.e. it would be very reasonable to assume that health insurers capture some of this lost ESI demand in the individual HIE market. Insurers are free to price their individual-market HIE offerings according to the associated risks; and the Bill provides for risk sharing corridors that protect insurers’ downside during the particularly risky first two years (please see an important correction from our Oct. 6th, 2009 call below). The only way employees buying individual coverage stay on HIEs long-term is if the mechanics of the exchanges are corrected to mitigate adverse selection; or a relatively low-cost public option is made available.
It follows that we think it’s a bit extreme to assume that households representing 13% of ESI premiums both enroll in HIE’s and are permanently lost as business to the commercial insurers (i.e. the commercial insurers either do not participate on the individual HIE’s; or, these households fall under an immediately available public option). Nevertheless we model this loss in our valuation scenarios, later in this call.
The Effect of ‘Cadillac-Plan’ Excise Taxes on Premium Growth
Turning to the high end of commercial insurers’ ESI business, we modeled the effect of excise taxes on high-premium or so-called ‘Cadillac’ plans. The current distribution of annual premiums for single (Exhibit 6) and family (Exhibit 7) coverage currently includes very small percentages of total enrollees that exceed the proposed 2013 excise tax thresholds of $8,000 for single and $21,000 for family coverage. High-risk professionals have higher limits, as do the 17 highest cost states for a limited period. For the sake of simplicity we ignore these higher limits, which produces a more conservative estimate of earnings power effects.
We assume that average premia in the individual and family markets continue to grow in line with historic (’99 – ’09) real growth in health costs (6.3%), multiplied by CBO’s current year-by-year CPI-U assumptions, and 1% population growth. We then model the percent of enrollees in either market having premiums that exceed the threshold in any year between 2013 and 2020 (Exhibits 8 and 9 for single and family coverage, respectively).
Recall that premiums in excess of the threshold are subject to a 40 percent excise tax payable by the insurer, or in the case of self-insured plans by the sponsor. The excise tax is applied only to the portion of premiums above the threshold, not to the entire premium.
For modeling purposes, we assume that the threshold is a brick wall – that the tax is steep enough that no one buys a plan with a premium above the threshold.
Thus over time, as premiums inflate in-line with underlying health costs, premiums stack up against the brick wall. Sometime after 2020, everyone is paying the maximum, and premium growth is limited to the rate at which the threshold grows: CPI-U + 1%. Along the way, as more and more enrollee’s hit the threshold (Exhibits 8,9), revenue growth falls from the product of health cost inflation and population growth, to the product of population growth and threshold growth (Exhibit 10). Plainly the excise tax threshold decelerates insurers’ top-line growth, but never forces it into a real decline.
It’s very much worth asking whether or not the excise tax truly changes insurers’ future revenue profile, or whether this deceleration would have happened on its own – our belief is that such a deceleration has been inevitable for some time for simple macro-economic reasons. Quite apart from the violent swings in enrollment inherent in economic down-cycles (Exhibit 11), we see fewer and fewer employees accepting the offer of ESI as premium costs grow relative to incomes (Exhibit 12). What’s happening is that as health costs grow faster than wage, the cash left over
after paying insurance premiums has declining purchasing power. For some households the health cost / wage relationship is no longer considered worthwhile (again, Exhibit 12); as health cost growth exceeds wage growth this ultimately becomes true for the population in general. Exhibit 13 revisits a simple calculation from our last call, wherein we showed that at current relative rates of health cost and wage growth, that household’s real non-health purchasing power begins falling – logarithmically – by roughly 2019. Thus in our view, the excise tax simply codifies a specific rate of decline in health premium growth relative to wage that will generally happen, irrespective of whether reform legislation passes.
We recognize the near- to mid-term effects of a falling job market on enrollment (Exhibit 11) and medical loss ratios, but believe that insurers’ relatively low valuations have more to do with longer term structural concerns. Accordingly we adopt a longer-term framework as context for valuation.
We compare eight strategic scenarios for health insurers’ earnings power to a standard assumption of long-term SP500 earnings power. For the SP500, we assume long-term earnings growth of 5.6 percent (the product of GDP (2.2%), CPI-U (2.0%), and productivity (1.0%)). For both SP500 and insurers’ earnings, we use a discount rate of 9.4 percent; this is equal to the rate of total returns on equities between 1926 and 2008 (10.43% at 3.04% inflation) adjusted for the forecast period’s lower anticipated inflation rate of 2 percent. For comparison we also show valuations under discount rate assumptions of 8 and 10 percent.
The eight scenarios for insurers’ earnings are arrangements of three ‘variable’ strategic assumptions set against a backdrop of three ‘fixed’ strategic assumptions. In all scenarios the ‘fixed’ assumptions apply; these are: insurers’ sales follow the declining trajectory modeled in Exhibit 10; revenue cannot grow faster than GDP in any year after 2020; and, insurers see no productivity gains at any point in the long-term comparison. The three variable assumptions are: whether or not insurers see a 13 percent drop in ESI sales in 2013, whether or not insurers see margin compression (to 80% of historic net margin levels), and whether or not insurers are affected by another health reform initiative at a later date (specifically modeled as a re-set of health care as a % of GDP to year 2000 (CPI-adjusted) levels; this re-set begins in 2020 (as health costs eliminate real purchasing power gains in fiscal and consumer settings) and is complete
by 2025). We use 2009 sales and average net margins for ’95-’08 as a starting point for insurers’ earnings. The earnings trajectories are shown graphically in Exhibit 14 (earnings are indexed to 2009 = 1, except for the scenarios in which insurers margins are reduced, in which case 2009 earnings = 0.80). The relative values of insurers’ earnings power relative to SP500 earnings power, for each scenario and under a range of discount rates, are shown in Exhibit 15. Present valuations are consistent with the most severe scenario, i.e. loosely speaking the group prices in a 13 percent reduction in revenues in 2013, a reduction in healthcare as a percent of GDP to year 2000 levels by 2025, and future net margins that are 20 percent below historic (’95 – ’08) averages.
It bears emphasizing that valuations attribute a zero probability to legislative failure (yet we see this as a very material possibility), zero productivity gains (despite nearly a decade of continued real pricing growth for insurers); and, zero probability of insurers participating in expanded demand for coverage as a result of expanded government subsidies (which on the assumption legislation passes, is equivalent to 100 percent odds of a 100 percent dominant public option in the HIE for individual policies). We conclude that insurers are undervalued.
We also note health insurers’ remarkably low valuation relative to other healthcare sub-sectors (Exhibit 16), and believe this has signal value with respect to the market’s point of view. If the market were suddenly pricing in the longer-term secular reality of falling health cost growth relative to wage growth, this reality almost certainly would be priced in across healthcare sub-sectors, since sales of the inputs to care have to fall just as much as premiums. This argues that the market is focused on reform-related effects to insurers, rather than the broader secular realities that affect all health care subsectors more or less equally. Given that we can only begin to justify current valuations on combined and aggressive assumptions regarding both reform-related and secular pressures, we’re further convinced that health insurers are undervalued.
- Our valuation framework is longer-term; over this time period we expect the annual tax on insurers to be a pass-through with little or no effect on earnings power, though we acknowledge that the tax creates uncertainties around 2010 earnings.
- We assume premiums rise and fall in proportion to wage.
- Medicaid eligibility extends to 133% FPL under the Senate Finance bill.
- Keep in mind that more efficient underwriting for small and medium employers in an HIE would work to shift this line to the left as the HIE’s emerged (’13 – ’15), but that health cost growth relative to wage would more than reverse that left shift over time (’15 and beyond).
- CORRECTION: In our October 6th call we incorrectly assumed that individual premiums in the HIE’s are permanently capped at a percentage of income, and that as premiums rise above these percents of income due to both health cost inflation and adverse selection, that the overage was entirely a claim on general revenues. 2013 premiums for individual coverage in the HIE’s are capped as a percent of income, however in subsequent years enrollees are liable for the same share of total premiums that they would have paid if premiums had hit their income limit in 2013. For example, if a family purchasing health coverage on an HIE in 2013 had spent to its limit of income, and in so doing paid for half of their total coverage cost, in 2014 and beyond they would continue to be responsible for half of total coverage cost as premium costs grew, irrespective of what percent of income this represented. Inevitably health cost inflation and adverse selection would make individual coverage in the HIE’s, as they are presently structured, less attractive than ESI coverage. See p. 27 of the amended Senate Finance bill, 2nd to last sentence in the first full paragraph. See also p. 2 of CBO’s 9/22/09 letter to Baucus.
- 13.8%; present levels are roughly 17% even if we eliminate recessionary effects