Consumers add Risk, Insurers add Power; Consequences for Providers and Suppliers
July 30, 2010
Consumers add Risk, Insurers add Power; Consequences for Providers and Suppliers
- As health costs grow faster than incomes, consumers likely take ever greater steps to reduce the growth of their health insurance premiums
- In the immediate term, this means taking more risk, i.e. accepting a higher deductible — high-deductible health plan (HDHP) premiums are on average 18% less than other forms
- HDHP’s do nothing to decelerate underlying price inflation, thus ultimately we believe consumers will be forced to the alternative – giving insurers greater power to choose products and services, and thus greater power to negotiate the input prices of care
- We fully expect both of these trends (greater risk to consumers, greater product and service selection power to insurers) to play out, and to do so sequentially (risk-shift first) with some overlap
- As consumers choose to bear more risk, unit growth falls. And, mix shifts away from discretionary products and services in outpatient settings, toward less discretionary care in in-patient settings. Private insurers’ (AET, CI, WLP, UNH) revenue growth slows, though we believe valuations provide more than adequate room, and that reform-related enrollment gains are a substantial offset. Retail-based prescription drugs — particularly for less symptomatic conditions – face perhaps the greatest relative risk. Inputs to in-patient care are less susceptible, though the nature of the underlying condition bears consideration – e.g. demand for relatively discretionary and/or interchangeable orthopedic devices is more at risk than demand for best-in-class chemotherapies
- As insurers gain power over product and service selection, price growth slows. Sub-sectors that are reliant on rapid price growth and/or whose product portfolios are heavily weighted with interchangeable products (e.g. large cap pharmaceuticals especially, medical devices to a lesser extent) are at greater relative risk; sole-source suppliers of products for more severe conditions face little incremental risk (e.g. biotechs). Health insurers that tend to have large shares of local market underwriting can build requisite networks and succeed (UNH, WLP); this is far more difficult for insurers with smaller local-market shares (AET, CI), who risk disintermediation by established local networks
Quite apart from the direct effects of federal health reforms, we see the nature of health insurance changing substantially in two fundamental ways: consumers are choosing to bear more risk; and, we believe consumers soon will choose increasingly restrictive plans (ceding more choice over providers, products and services to insurers).
The common driver of these two changes is the excess rate of health cost growth over wage growth. We remain convinced that health cost growth will continue to exceed wage growth for the foreseeable future, the effect on consumers being that health insurance premiums – assuming no change in the nature of insurance purchased – command an ever-larger portion of household income. At present rates of health cost and wage growth, the average American household’s real non-health purchasing power begins falling – exponentially — in roughly 9 years (Exhibit 1).
Plainly consumers (voters) are unlikely to sit still for this, thus our long-standing conclusion that we’re headed toward a further round of reforms that take place in, or on the threshold of, a health-cost driven economic crisis.
Trading risk and choice for lower costs …
The focus of this note is what happens along the way to some eventual structural reckoning … as insurance premiums ‘try’ to consume an ever larger portion of total household income, households inevitably resist losing their non-health purchasing power. Demand for health insurance is substantially more elastic than demand for other healthcare products and services, so we should naturally expect consumers to buy ‘less’ healthcare insurance as its price rises relative to incomes. For the moment, the most available means of reducing health insurance premiums is to assume more risk – e.g. to choose a high-deductible health plan (HDHP) instead of a more generous alternative such as a preferred provider organization (PPO). Ultimately we believe that the shift to HDHP’s does nothing to slow underlying health cost inflation, as the vast majority of health spending (and thus inflation) occurs at levels above (and thus unaffected by) even HDHP deductibles (Exhibit 2).
Ultimately this leads households to the remaining (and at present, not commonly available) alternative, that of giving health plans sufficient negotiating leverage over product and service providers to reduce the input costs of care. More simply, this means eventually choosing to enroll in much more restrictive networks than are commonly chosen today.
Setting the stage: How we got to where we are today, and why the current ‘trade’ is to take more risk as opposed to giving up choice …
Exhibit 3 shows US market shares of various plan designs since 1988; for simplicity similar plan designs (less-restrictive indemnity and point-of-service (POS) plans are combined, as are more restrictive PPO’s and health maintenance organizations (HMO’s)). In very broad terms the period 1998 to 2002 was characterized by a shift from very loosely organized networks of care (indemnity or POS) to more tightly organized networks (HMO or PPO). Thus the big picture until 2002 was one of health insurers shifting from little or no power to select products and services for beneficiaries, to at least some level of power over what products and services are made available. Since 2002 the market shares of more traditional versions (HMO/PPO) of restrictive plan designs have been relatively static, and the major enrollment share trend has become the uptake of HDHP’s – though to be clear, HDHP’s are effectively PPO’s with high deductibles, thus the broad march from very permissive indemnity/POS plan designs to at least somewhat restrictive HMO/PPO plan designs hasn’t really slowed – we’ve just added the feature of greater consumer risk (higher deductibles with HDHP’s) to the ever more dominant PPO model.
At this very general level one might conclude that the shift from free choice (indemnity / POS) to restricted choice (PPO/HMO) is largely complete, but this general view belies a deeper trend – even though some degree of restrictive network has become the norm, beneficiaries have shied away from more tightly managed networks (HMO’s), instead preferring more freedom of choice (PPO’s). Since the late 90’s a falling percentage of persons offered the most restrictive choice – an HMO – actually take that option (Exhibit 4). Rather, beneficiaries – when given the choice – have come to increasingly prefer the less restrictive PPO option. We conclude that consumers still have plenty of (eventual) room to trade freedom of choice for lower premiums.
As a quick aside, at least part of what’s behind consumers’ preference for PPO’s over HMO’s appears to be a degree of mis- pricing. Beneficiaries’ annual premium costs in PPO’s and HMO’s are now quite similar, after years of HMO’s having been the lower cost option (Exhibit 5). This appears to have happened for at least two reasons: First, employers’ absolute dollar subsidies of health insurance coverage tend to be larger as premiums grow larger – so beneficiaries have an incentive to choose more expensive plans, in order to receive a more generous employer subsidy. Conversely, and more to the point — if beneficiaries choose cheaper plans, they don’t get enough of the net savings to make it worth their while to give up the requisite control over their health care choices. Second, as HMOs’ market share of total beneficiaries has fallen, presumably so has their negotiating leverage over providers, thus HMO’s now arguably have much less of an operating economic advantage over other plan designs (such as PPO’s) than in the past.
So if consumers wanted to trade freedom of choice for lower premiums – as we suspect they ultimately will – at the moment they’re simply not ‘paid’ to do so. In contrast, consumers can lower their premiums (about 18%) by accepting more risk (enrolling in an HDHP; Exhibit 5, again). Thus at least for the moment, consumers seeking lower premiums have little choice but to accept greater health cost risk.
What happens when consumers bear more risk?
From the RAND Health Insurance Experiment (HIE) we know generally that beneficiaries paying a higher proportion of medical costs out-of-pocket tend to consume less healthcare; and, that this is the result of fewer – as opposed to less intensive – medical encounters. More simply, patients with high out-of-pocket cost shares appeared more selective about whether or not to interact with the health system, but once in the system appeared to have little if any influence over what then ensued.
The RAND HIE compared a number of plan designs, including four that varied simply by the level of co-insurance beneficiaries had to pay for any type of healthcare – free, and 25%, 50%, or 95% co-insurance. In all cases out- of-pocket spending was capped at either an absolute dollar figure, or a percentage of household income. In the HIE, as beneficiaries bore a greater proportion of their health costs out-of-pocket (i.e., as co-insurance rates went up), total health spending fell (Exhibit 6). In-patient spending fell less quickly than outpatient spending (also Exhibit 6), which implies that the elasticity of demand for in-patient services was lower than for outpatient services. Arc elasticities are consistent with this conclusion; across co-insurance ranges demand for care in the HIE was more elastic in the outpatient setting (Exhibit 7). We might reasonably infer that demand for in-patient care is more inelastic because typical in-patients tend to be sicker – and thus in more immediate and unavoidable need of care – than typical outpatients – or, more to the point, that sicker patients’ demand is less elastic. HIE results are consistent with this interpretation; unsurprisingly the amount of health care expenditure in HIE tended to vary fairly substantially according to health status (Exhibit 8).
Turning to a more contemporary dataset (Medical Expenditure Panel Survey (MEPS), 2006) we analyzed health expenditures as a function of the proportion of health spending that private insurance beneficiaries bore out-of-pocket, and found that a 10 percent increase in out-of-pocket share appears to result in roughly a 5 percent decline in total health expenditure (Exhibit 9). In very rough terms, this broad indicator of elasticity is reasonably consistent with RAND elasticity measures (wherein a 50% co-insurance rate corresponds to a 24% reduction in total expenditures (Exhibit 6)).
Various others have looked at how large deductibles affect demand, and of particular interest to us as investors are studies that attempt to find relative changes in demand across discrete healthcare sub-sectors. Consistent with the general trend in the RAND HIE, Parente et al found that demand for outpatient services was more elastic than demand for in-patient services; unlike RAND, Parente breaks out prescription drugs separately, and appears to show that prescription drug demand is more elastic than demand for in-patient services (Exhibit 10). Others have shown that consumers with higher out-of-pocket costs may consume fewer prescriptions; and in our own previous work we’ve shown that prescription drugs for asymptomatic conditions (e.g. hypertension, hypercholesterolemia) are more susceptible to out-of-pocket costs than drugs for either symptomatic (e.g. allergy) or life-threatening (e.g. HIV) conditions.
The obvious and simple pattern is that people who are more exposed to the costs of their healthcare consumption tend to reduce their spending where and when (and if) they have discretion to do so. It follows that companies selling to patients whose consumption is less discretionary are better positioned.
Broadly speaking drug sales should come under pressure simply because of the direct out-of-pocket cost that retail drugs represent to patients that have not yet met their higher deductibles, though the nature of the underlying condition remains key. Crestor (the 12th largest US product in 2009, AZN), as a treatment for an asymptomatic condition is less secure than Nexium (the 2nd largest US product in 2009, also AZN), which treats a more symptomatic condition, though both products are susceptible to being replaced by less expensive therapeutic equivalents (e.g. simvastatin and omeprazole, respectively). Retail drugs for more symptomatic and/or life-threatening conditions (e.g. anti-psychotics, anti-depressants, insulin, HIV anti-virals), particularly when choice of drug is highly controlled by the prescribing physician, are comparatively protected. Drugs purchased directly by physicians (e.g. oncolytics) or institutions (e.g. anesthetics, anti-biotics) presumably are even less susceptible to high deductibles, as product selection is generally if not completely out of the hands of the patient, and patients in these settings tend to have more severe conditions.
Suppliers to hospitals of ‘general’ inputs to care (e.g. Alcon, Baxter, Becton-Dickinson, Covidien, CR Bard, Thermo-Fisher and to some degree Abbott) also would appear to be relatively protected from the elasticity effect of higher deductibles, simply because they are selling into the less elastic inpatient setting. Suppliers of specialized / innovative inputs to hospitals arguably – like pharmaceuticals – still should be evaluated on the elasticity characteristics of the specific underlying condition(s) treated. For example, we would expect suppliers of devices and technologies for cardiac intervention (e.g. Boston Scientific, Heartware, Thoratec, and to some extent Medtronic) to be less affected than suppliers whose product lines may be used in at least somewhat more discretionary orthopedic procedures (e.g. Zimmer, Smith and Nephew).
Plainly a shift to greater risk-bearing is an incremental negative for health insurers focused on the employer-sponsored market (Cigna, Aetna, United Healthcare, WellPoint); since 2006, HDHP premiums have been on average 18% lower than PPO premiums. For the moment, we acknowledge that HDHP plans may have higher gross margin percentages because of favorable selection, but emphasize that favorable selection wears off as HDHP’s become a larger proportion of total underwriting; and, that the emergence of MLR restrictions may tend to normalize MLR trends across plan types. Nevertheless we see HDHP’s as a means of creating health coverage options that are ‘salable’ to consumers who inevitably will try to reduce premium costs relative to income, and continue to believe that insurers’ valuations are unrealistically low.
What happens when consumers give up (some) freedom of choice?
As opposed to shifting risk to beneficiaries through higher deductibles and/or co-pays, HMO’s seek to control costs by exercising greater decision-making power over beneficiaries’ health care consumption. In effect, beneficiaries give HMO’s some level of proxy over their health care choices, which HMO’s can use as negotiating leverage over providers.
As we’ve shown, HMO’s are substantially less popular than before, though their years of peak influence serve as a natural experiment that helps to show the downstream effects of more restrictive plan designs. Going on a combination of memory and data, we loosely – and somewhat arbitrarily — set the years of HMOs’ peak influence as 1990 – 2000. HMO’s as a percent of total employer-sponsored enrollees passed through a critical mass level of 20% in the very early 90’s, peaked at around a third of enrollees in the mid-90’s, and had fallen well below a quarter of enrollees by 2000. HMO’s share of total enrollment has continued to decline since (Exhibit 11).
Recognizing that HMO’s were a major – but by no means the only – influence during this period, we examined components of US healthcare demand and demand growth during the relevant timeframe. Beginning with a breakdown into mix, inflation, and per-capita unit intensity, the major change in healthcare demand from 1990 – 2000 was a deceleration of healthcare price growth (Exhibit 12). Health price changes are broken down by major components (hospitals, physicians and prescription drugs) and expressed in real terms in Exhibit 13. At the mid-90’s peak of HMO influence, real healthcare price growth was at its nadir; this is at the very least strongly suggestive that the HMO model effectively reduces underlying rates of growth in health prices, presumably because of the greater negotiating leverage HMO’s hold over providers by virtue of their ability to influence which products and services beneficiaries do (or do not) consume. Our take on utilization during the HMO era (Exhibit 14) is that HMO policies simply built on pre-existing Medicare policy with respect to (rapidly falling) hospital lengths of stay, had little effect on number of physician visits (but did reduce the specialty-intensity of those visits – not shown), and somewhat ironically increased prescription drug utilization by virtue of adding credit-card like elasticity effects to retail drug transactions.
We believe that the hospital length-of-stay and drug utilization effects of HMO’s during the ‘90’s are one-time effects that would not be repeated. We do however believe that HMO’s effect on real price growth – and arguably on mix as well — would return if and when consumers chose to lower premiums by enrolling in more restrictive networks. Turning to investment implications, it’s important to bear in mind that HMO type negotiating pressures are some ways off. As convinced as we are that consumers ultimately will choose to sacrifice product and service choice in exchange for lower premiums, we again emphasize that they cannot do so at present – too few consumers are presently willing to make that trade, and weakened HMO networks generally cannot produce the necessary savings to deliver meaningfully lower premiums.
At the risk of over-simplifying and over-generalizing, HMO’s work by exercising their (far more substantial) power to choose among various suitable alternatives – largely based on cost / price. E.g. generic omeprazole instead of branded Nexium, Crestor at a low price as opposed to Lipitor at a higher price, this knee implant on contract as opposed to that knee implant which is not on contract. It follows that industries having higher-priced products and services that are on some level interchangeable or substitutable are most at-risk. Traditional large-cap pharmaceuticals (Pfizer, Merck, Astra-Zeneca, Bristol-Myers, Eli Lilly, Sanofi-Aventis, Johnson & Johnson, Novartis, GlaxoSmithKline, Roche) are at risk to the extent that they continue to rely on rapid real rates of price growth, and parallel exploitation of basic discoveries which tends to lead to multiple competing versions of highly interchangeable products. Manufacturers of innovative but potentially interchangeable medical devices (e.g. Stryker, Boston Scientific, Medtronic , St. Jude, Zimmer, Smith & Nephew) might finally face pricing pressure sufficient to overcome the product selection sovereignty of cardiovascular and orthopedic surgeons. In contrast, sole-source suppliers of non-interchangeable products for severe conditions would remain relatively less exposed to pricing pressures (e.g. Amgen, Biogen, Celgene, Genzyme, Gilead).
Whether large-cap insurers benefit from a shift toward more tightly managed networks to our minds very much depends on their local market shares. Insurers that tend to have very high shares of total underwriting in the markets they serve (e.g. United Healthcare, WellPoint) arguably would be in a strong position to develop the requisite networks; insurers with relatively low shares of local market underwriting arguably would face an increased risk of disintermediation (Aetna, Cigna).
- Importantly, the fiscal picture is quite similar – at current tax and borrowing levels, real federal spending on non-health items begins falling around 2019. ↑
- Parente, et al. “Evaluation of the Effect of a Consumer-Driven Health Plan on Medical Care Expenditures and Utilization” Health Services Research 39:4, Part II (Aug 2004) pp 1189 – 1210 ↑
- Our interpretation of Parente’s table is narrowed to the first year of comparison. We show the subsequent years for sake of completeness, but emphasize that the rapid rate of growth in health spending by CDHP (i.e., high-deductible) enrollees in Parente is un-explained, and does not appear to be consistent with other evidence ↑
- E.g. Greene et al. “The Impact of Consumer-Directed Health Plans on Prescription Drug Use” Health Affairs, Volume 27, Number 4, pp 1111-1119 ↑
- Under previous indemnity plans, patients would pay full price at pharmacy out-of-pocket and seek re-imbursement; the advent of prescription benefit ‘cards’ effectively eliminated the cash flow effect, and at least in our interpretation accelerated retail consumption ↑