US Healthcare Demand, Part 1: ‘Baseline’ Growth

Print Friendly, PDF & Email

Richard Evans / Scott Hinds / Ryan Baum


203.901.1631 /.1632 / .1627 richard@ / hinds@ /


September 4, 2012

US Healthcare Demand, Part 1: ‘Baseline’ Growth

  • This note quantifies a ‘baseline’ rate of real growth in US healthcare demand, which we define as the rate of growth one would expect sans any upcoming secular, cyclical, or reform effects
  • We estimate baseline growth of 4.8% +/- 0.9% through 2021. This consists of 80bp population growth, 70bp population aging effect, 2.1% increase in age-adjusted per capita utilization, and 1.2% real pricing
  • This 4.8% estimate of forward real demand growth is 80bp higher than occurred since 1990, but on par with real demand growth since 1970. Most recent (2Q12) real demand growth (for health services only) was just 1.8%, well below historic averages and our forward ‘baseline’ estimate. The current (2Q12) growth rate in age-adjusted per capita utilization (1.2%) is almost half the ‘normal’ or ‘baseline’ rate, reflecting high rates of unemployment (and thus low penetration of employer-sponsored health insurance)
  • Healthcare trades at a relative (trailing) PE of 0.95x the broader SP500; below the longer-term trend of 1.06x. If the market genuinely expected our baseline of 4.8% real demand growth the relative multiple should be >1.0x. This suggests the market believes the net impact of secular, cyclical, and/or reform effects will be to reduce demand growth below the baseline rate
  • Judging by the pattern of valuations across healthcare sub-sectors, the market has confidence in underlying unit demand, but skepticism regarding future contributions of innovation and pricing
  • Non-Rx Consumables (e.g. BCR, BDX, OMI, CFN) and Diagnostic Laboratories are far more reliant on underlying volume demand than on innovation or real pricing power; these sub-sectors trade at parity to the SP500, which implies market faith in underlying demand volumes
  • Hospitals and Commercial HMOs trade at steep discounts to the SP500 (0.7x and 0.8x, respectively); given the market’s apparent expectation of strong unit demand, the discount almost certainly reflects non-volume related concerns. Hospital valuations imply negative real pricing; this assumption is too severe, and we see Hospitals as a value. Commercial HMOs imply some combination of rising MLRs near-term and disintermediation middle- to longer-term; we expect stable MLRs and see the Commercial HMOs as a value in front of the reform roll-out; however after the roll-out of reforms disintermediation pressures are likely to build steadily if employers shift employees to health insurance exchanges
  • Major Pharmaceuticals trade at a discount to healthcare and the SP500, implying either or a combination of two scenarios: continued mix erosion even after the current generic ‘wave’; and/or lost real pricing power. Both of these pressures are likely, so the discount makes complete sense. In contrast Biotech trades at a premium; we believe this fairly reflects longer product lifecycles and more sustainable pricing power than is the case with more traditional drugs. Current valuations give Biotech only limited credit for probable levels of innovation and pricing power, and if anything appear too low

This note seeks to answer a straightforward question: If we ignore secular, cyclical and reform effects, what is the ‘inherent’ or ‘baseline’ real rate of growth in US healthcare demand? Having established a baseline, three follow-on notes will dimension the likely effects of secular (e.g. threats to real pricing power, slow erosion of employer-sponsored health insurance, the trend toward higher-deductible health plans), cyclical (primarily the shifting of households into and out of employer-sponsored insurance), and reform effects (e.g. size and timing of the Medicaid expansion, uptake rates of exchange-based insurance among persons currently uninsured, and possible effects of shifting employees onto the health insurance exchanges)

Demand = People x Utilization x Price

At its simplest level, real healthcare demand is the product of population growth, changes in per capita utilization (units and mix), and real pricing. Since 1970 the average product of these three growth rates (i.e. total real demand growth) is 4.7% (Exhibit 1). Growth in per capita utilization has been the largest[1] (and most volatile) driver by far, followed by real pricing and population growth

Per capita utilization growth (historic average = 2.4%) can be unpacked further into its three components: aging effects (0.3%), plus the effects of innovation and intensity (which together equal 2.1%). Innovation drives utilization growth in two ways: 1) treatment options become available for persons that previously had no option, so the average amount of care consumed per person rises; and 2) newer (and generally more expensive) treatment options displace older treatment options, raising the average real cost of care consumed per person. We define intensity as changes in the amount of care consumed for reasons independent of pricing or innovation; the most powerful influence on age-adjusted intensity is the availability and average generosity of health insurance. It’s impossible to say precisely what portion of age-adjusted per capita utilization growth is driven by innovation or intensity, but we can get a rough estimate by backing per capita volumes and age effects out of available measures of per capita intensity. These analyses very roughly suggest that, with the exception of hospitals[2], innovation and intensity have had similarly sized effects on demand growth (Exhibit 2)

Real pricing effects vary quite a bit across time, and follow a different pattern than changes in utilization (Exhibit 3). Changes to age-adjusted utilization are largely explained by the employment cycle (Exhibit 4, more on this in Part 3); in contrast real pricing power is the natural consequence of for-profit suppliers operating in an environment of very low average elasticities

Population growth is the final part of the growth equation, and is both small and predictable, averaging 1.0% since 1970, with an SD of just 40bps

What’s a Reliable ‘Base’ of Demand?

The demographic components of demand growth (population growth, aging) are highly predictable (and thus reliable); changes in utilization and real pricing power are much less predictable

Total demographic effect since 1970 has averaged 1.3%; current US Census Bureau projections imply a demographic effect of 1.4% from 2011 – 2021 (the product of 0.8% population growth and 0.7% aging effect) (Exhibit 5). As a point of reference, total 2011 real demand growth was only 2.2%, just 90bps above the underlying demographic growth rate

Off of this highly predictable demographic base, the matter of reliable total growth is a question of how much growth can reasonably be expected from the combined effects of real pricing and changes in age adjusted per capita utilization. Importantly, pricing and utilization tend to move independently of one another (Exhibit 4, again), meaning we’re inherently more likely to experience the combined weighted average of these two values than a simultaneous combination of extreme highs or lows in either value. The average product of age adjusted per capita utilization and pricing power since 1970 is 3.4%, with an SD around the annual growth rate of 2.2%, an SD around the 5 year average growth rate of 1.3%, and an SD around the 10 year rate of 0.9%. Simplistically, ignoring the likelihood of secular, cyclical or reform effects, the forward expectation of demographic demand growth (1.4%) plus the long-term average of utilization and pricing (3.4%) produce a reasonable estimate of ‘baseline’ real growth (4.8%) over the middle-term. Still ignoring any specific assumptions regarding secular, cyclical, or reform effects, we might reasonably expect +/- 0.9% variance around this 4.8% ‘baseline’ over the next decade

To be clear, we’re not suggesting random sources of upside / downside to the baseline 4.8% real growth estimate. In practice, risks to ‘baseline’ growth have been the byproduct of either large scale policy changes or extreme economic circumstances. Since 1970, the product of age-adjusted per capita utilization and real pricing has fallen more than 1 s.d. below the 1965-2011 average on 5 occasions: 1) hyperinflation (CPI-U = 11.0%) in 1974; 2) hyperinflation (CPI-U = 13.5%) and falling employment in 1980; 3) sharp declines in hospital utilization in response to policy changes by Medicare and commercial payors, coupled with the effect of tightly managed commercial HMOs on real pricing (peak combined utilization X price headwind occurred in 1996); 4) the 2001 recession; and 5) the post-Lehman period

What’s happening now …

Real demand growth from 2007 to 2011 has been only 2.5%, around half the ‘baseline’ real growth rate, and only 1.2% above underlying demographic growth. Most of what’s ‘missing’ from the 2007 to 2011 growth equation is per capita utilization, which grew 1.3% v. the long-term (1970 – 2007) average of 2.5%; real pricing grew 0.4% v. the 1.3% long term average. Taken together, ‘lost’ utilization and pricing explain 89% of the post-Lehman period’s slowing of real healthcare demand relative to the long-term average. More on this in Part 3, but practically all of the lost utilization can be explained by employment losses, which presumably impair demand as a consequence of households losing employer-sponsored health insurance

To measure recent quarterly demand we have to narrow our perspective to health services demand, as these are the only data available for recent quarters. After bottoming in April of 2010, health services demand rallied briefly and has subsequently slowed; we believe this pattern is the byproduct of underlying modest jobs growth off of the employment trough, followed by more recent slowing jobs growth (Exhibit 6). 2Q12 real health services demand growth was 1.8% (2.0% utilization, -0.2% real pricing); we expect 3Q12 real growth of 2.3% (2.0% utilization and 0.2% real pricing). Note these utilization figures include population growth and aging effects, meaning utilization growth is only about 70 bp above the underlying demographic baseline of about 1.3%

What’s implied by current valuations

Since 1991, the SP500 HC Sector Index has traded at an average trailing relative PE (to the SP500) of 1.06, with a range of 0.71 (2009) to 1.55 (2000) (Exhibit 7). If our baseline demand expectation of 4.8% real growth were to be realized, we would have every reason to believe healthcare as a broad industrial category would warrant at least this relative multiple going forward[3]. The current relative PE of 0.95x implies at least slight skepticism that our demand baseline is realistic

The question is why, and the answer appears to be less about the strength of underlying unit demand, and more about innovation, pricing power, and disintermediation risks

Non -Rx Consumables[4] and Diagnostic Laboratories trade at parity to the SP500 on forward earnings (Exhibit 8); because these subsectors have relatively little exposure to the revenue effects of innovation or real pricing power, this implies the market expects ‘baseline’ levels of per capita unit demand to normalize[5]

In contrast, Major Pharmaceuticals trade at a +/- 10% discount to the SP500 on forward earnings (Exhibit 9)– which implies that these names will experience less than ‘baseline’ demand growth, despite being research based businesses with substantial real pricing power. Because the volume-bellwethers’ (Non-Rx Consumables and Diagnostic Laboratories) valuations imply very nearly baseline levels of growth, this argues the market expects pharma to fall short of baseline demand growth for reasons unrelated to underlying volume demand – and these can only be some combination of mix erosion and reduced real pricing power. Because returns to R&D spending are negative, and because of mid-term threats to pricing power, we believe pharmaceutical companies will in fact produce revenue growth below the rate of baseline demand growth, meaning some level of discount to both the SP500 and overall healthcare makes great sense

Commercial HMOs and Hospitals are quite dramatically cheaper than the SP500 on forward earnings (Exhibit 10). Again, because the volume bellwethers are at parity, this suggests the market sees problems with these subsectors that are unrelated to underlying volume demand. In the case of Commercial HMOs, because employment gains are more beneficial than for other healthcare subsectors[6], and because employment is more likely to rise off current near-trough levels than to fall further, the fact that forward valuations are at such low levels implies rather extreme concerns about either or some combination of three threats: falling MLRs, falling average contract values, and/or disintermediation. Longer-term, we see disintermediation risks and falling average contract sizes as employers shift employees to health insurance exchanges; however in the near-term we see improving fundamentals in an (eventual) jobs recovery, where the market seems to expect MLR contraction on price competition. Accordingly we see the Commercial HMOs as a value ahead of the reform roll-out

Unlike Commercial HMOs, Hospitals really cannot be disintermediated, so the fact these stocks discount demand growth well below the 4.8% baseline can only be explained by some fairly dramatic assumptions regarding Medicare / Medicaid rate cuts and/or rising collections headwinds. To justify current valuations, hospitals would have to lose all of their real pricing power, and a third of the benefit they typically see from rising per capita intensity. This seems overly pessimistic, as Hospitals appear to have sufficient commercial pricing power to offset Medicare / Medicaid rate cuts ahead of the full roll-out of healthcare reforms. Moreover, despite expecting collections headwinds[7] after the roll-out of reforms, and we can barely imagine a ten-year forward scenario with negative real pricing on a weighted average basis across all payors. Thus as is the case with Commercial HMOs, we see Hospitals as a value, particularly ahead of the reform roll-out

Biotech is the only other major subsector with remarkable valuation relative to healthcare overall, and as usual Biotech is the bright spot in terms of growth expectations (Exhibit 11). If anything valuations appear modest. The +/- 15% premium to the SP500 on forward earnings implies slightly better than baseline demand growth; all it takes to justify this valuation is for Biotech to experience the same pricing power as healthcare on average, plus slightly better per capita intensity growth (about 2.5% for Biotech v. 2.1% for healthcare overall). Because follow-on biologics are unlikely to produce the same types of generic headwinds as occur in big pharma, and also because new product pricing power in specialty categories appears sustainable over the medium term, we see upside in Biotech valuations

  1. Real pricing is an underestimate of the ‘true’ real-pricing value, and utilization is an underestimate of the ‘true’ utilization value. Real pricing pops-up in two places – one captured by the available indices, and one that is not (reliably) captured. Real change in the price of a specific and narrowly defined product used in healthcare (e.g. 50mg of drug X) is easily captured in the available pricing indices. In contrast, the tendency of new technologies to enter the market at prices that may exceed the economic value of their benefits v. older technologies is an effect that falls into the utilization calculation – but is in large part a price effect. Assume for example that disease A is treated by drug B, and that drug B costs $100. Drug C is then launched at $200, replacing drug B as the standard of care. The $100 difference in price between B and C will fall into the ‘utilization’ measure. However if the true economic value of drug C is only $50 greater than drug B, then in reality the market has ‘seen’ a $50 increase in utilization, and a $50 increase in real price (the amount by which C’s price exceeds its value)
  2. Average hospital days / capita fell by half between 1981 and 2001 as a result of changes in payment policy, rather than as the result of changes in the magnitude and nature of underlying demand
  3. In reality, our 4.8% baseline real demand growth assumption calls for an acceleration of real demand relative to the 1991 – present period, during which real demand growth was closer to 4.0%. Three major effects have to be considered – the 1990’s saw an enormous reduction in days of hospital care per capita, which limited overall per capita intensity growth in the period; pharmaceuticals pricing and product flow during the period contributed more to real demand growth than is likely to be the case going forward; and, per capita intensity fell to its lowest level in more than 50 years as employment troughed post-Lehman. Taken together, the two headwinds to demand growth (falling hospital days, recession) are larger than the tailwind of pharmaceuticals pricing and product flow, meaning actual 1991-2011 demand growth is less than would have been expected. The final (small) point is that total demographic effect over the next decade should be about 10bp greater than the 1991 – 2011 effect
  5. We might reasonably expect the ‘equilibrium’ relative multiple for these names to be somewhat below market, given their relative lack of innovation and pricing power; however because we’re starting from a near-trough in employment, these names should see more volume-driven operating gains than their peers as employment normalizes
  6. Enrollment growth obviously means rising revenues and the related potential for operating leverage; less understood but equally important is that the marginal enrollee improves the overall risk book, since she pays the same premiums as persons who have been employed all along, but her healthcare consumption takes around 14 months to rise to the average level of other employees
  7. We expect rising cheaper average policies among the employed-insured lead to higher patient out-of-pocket obligations, which translate into growing collections headwinds
Print Friendly, PDF & Email