US Healthcare Demand Part 2: Secular Headwinds

ginna
Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

Richard Evans / Scott Hinds / Ryan Baum

SEE LAST PAGE OF THIS REPORT FOR IMPORTANT DISCLOSURES

203.901.1631 /.1632 / .1627

https://twitter.com/images/resources/twitter-bird-blue-on-white.png richard@ / hinds@ / baum@sector-sovereign.com

@SecSovHealth

September 17, 2012

US Healthcare Demand Part 2: Secular Headwinds

  • We frame real growth in US healthcare demand as the simple product of: growth in persons, growth in per capita utilization, and growth in price. Our ‘baseline’ estimate of real demand growth, defined as the rate of growth one would expect in the absence of secular, cyclical, or reform effects, is 4.8%; this consists of 80bp of population growth, 2.8% growth in per capita utilization, and 1.2% real price growth
  • The 2.8% baseline estimate of per capita utilization growth breaks down further into 70bp of aging effect, with the balance split evenly between per capita intensity (units of care), and per capita mix (technical sophistication of a unit of care)
  • We anticipate three secular headwinds that in aggregate reduce baseline demand growth by approximately 90bp; all of these effects fall into the category of growth in per capita utilization
  • Health cost growth in excess of wage growth has driven employees to either refuse employer-sponsored insurance, or to choose cheaper forms of insurance. We expect these two secular effects to continue; together these reduce baseline demand growth by approximately 35bp
  • Declining real growth in R&D spending must reduce the contribution of mix to real demand growth, unless R&D productivity is rising. If anything R&D productivity appears to be falling, increasing the likelihood that innovation will contribute less to future growth. On a trailing basis, mix growth accounts for roughly 1.1% annual growth in real per capita demand; we expect this rate to fall by at least half, or 55bp

Summary and Conclusions

We believe secular effects in total are likely to reduce our 4.8% baseline estimate of real demand growth by approximately 90bp (Exhibit 1). There are 3 secular headwinds, all of which fall under per capita utilization. Because real price growth is likely to continue, premium growth is likely to exceed wage growth, which produces two of the three secular effects: declining uptake rates of employer sponsored health insurance, and the selection of cheaper forms of insurance (e.g. high-deductible health plans, or HDHPs). Combined, these secular trends reduce growth in per capita intensity (unit demand) by roughly 35bp annually. The third secular effect is declining growth in mix (the technical sophistication of a ‘unit’ of care); at a minimum this mirrors declining real growth in R&D spending, and we believe the effect of slower R&D spending growth is being magnified by declining R&D productivity. On a trailing basis, per capita mix growth accounts for roughly 1.1% annual growth in real demand; going forward we see the contribution of mix falling by roughly one-half, to 55bp

Details

In Part 1 of this series, we estimated ‘baseline’ real growth in US healthcare demand, defined as the rate of growth one would expect in the absence of secular, cyclical, or reform effects. Our estimate of real baseline growth is 4.8% +/- 0.9% over a ten-year period; this consists of 80bp population growth, 70bp population aging effect, 2.1% increase in age-adjusted per capita utilization, and 1.2% growth in real pricing

This note is Part 2 in the series, focusing on secular growth effects. We’ll use the same basic growth equation used in Part 1, specifically:

real growth = growth in persons X growth in utilization per-person X real pricing

Because the rate of population growth is reasonably constant, meaningful secular effects are limited to changes in per capita utilization, and changes in real pricing power

Secular pressures on utilization

Recall that per capita utilization itself consists of three variables: population aging, age-adjusted intensity, and mix. Population aging effects are straightforward; because health consumption varies greatly by age, as the blend of ages in the population changes, per capita demand accelerates or decelerates as a result. ‘Intensity’ refers to changes in per capita demand for ‘units’ of care, e.g. changes in doctor visits, hospital admissions, prescriptions and so forth. ‘Mix[1]’ refers to the technical sophistication of the average unit of care; a switch from a generic to a brand is a positive mix shift, and vice versa

Exhibit 2 shows the time series of aging effect on US real demand per- capita; we estimate that aging adds approximately 45bp of per capita demand annually as of 2012, and that the effect will grow to a peak of approximately 75bp by 2020. At the margin the effect is quite small – aging adds only 30bp of per capita real demand growth over the coming decade

On an age-adjusted basis, per capita utilization accounts for 2.1% of our 4.8% baseline demand expectation; in rough terms we believe intensity and mix each account for half of baseline utilization growth

Starting with secular effects on intensity: among the insured, the price and income elasticities of demand for healthcare are very low; because of this the main secular driver of age-adjusted intensity is the availability and average generosity of health insurance. Exhibit 3 summarizes health insurance sources for the US population since 1987; government sources of insurance (particularly Medicare and Medicaid) have become more predominant, employer sponsored insurance (ESI) has become less predominant, and rates of uninsurance have been steadily rising. On an age- and health-adjusted basis, a household with ESI will consume more than a household with government health insurance, and a household with government insurance will consume more than a household that is uninsured, thus the trailing effect of availability and source of insurance has been to limit real demand growth. Looking forward, we expect the roll-out of health reform to change Exhibit 3 fairly dramatically; specifically we expect substantially more households in government sponsored coverage, substantially fewer in ESI, and somewhat fewer uninsured. We’ll deal with these reform effects more fully in Part 4 of this series; for the meantime we’ll focus more narrowly on secular changes to the source and average generosity of health insurance, which lie predominantly within ESI

Fewer eligible workers are accepting their employer’s offer of ESI, and we believe this explains most of the secular decline in ESI as a source of health insurance (Exhibit 4). As health costs – and employees’ share of health costs – grow faster than the rest of the economy (and wages, Exhibits 5,6), employees’ non-health wage growth is effectively reduced; as this trend continues an increasing percentage of ESI-eligible households will simply choose to be uninsured. As an aside, there is an underlying cyclicality to employees’ share of cost growth – employers tend to shift more cost growth to employees during high unemployment, and absorb more cost growth during periods of low unemployment (Exhibit 7). Since 1987, the steady secular decline in the percentage of the population covered by ESI is responsible for a roughly 15bp annual headwind to demand growth[2]

An obvious alternative to not having health insurance is to simply buy less health insurance; in most markets the only valid option[3] is to reduce premium costs by accepting more risk, i.e. by enrolling in an HDHP[4]. HDHPs have grown steadily as a percentage of ESI over the last half-decade, and we expect this trend to continue as more and more households search for ways to lower premium costs. The effect of HDHP enrollment on healthcare consumption (a 14% decline, Exhibit 8) in the period immediately following a household’s switch from a more generous form of insurance is reasonably well established; however the longer-term effects are less clear. Because the majority of health costs are borne by a minority of patients (Exhibit 9) – and the great majority of these patients’ spending is beyond their HDHP deductibles – we believe the effect of HDHP enrollment on households’ consumption is much lower than 14% after the first year. Nevertheless if we make the conservative assumption that the drop in consumption seen in the first year of HDHP enrollment persists, then HDHP enrollment growth over the last six years has produced an average annual demand headwind of 20bp (Exhibit 8, again)

Shifting to mix effects – if we momentarily assume that R&D productivity is constant, then mix gains from R&D spending are a simple matter of how much R&D is spent. In absolute terms, since 1997 real growth in healthcare-related[5] R&D spending has been decelerating, particularly recently (Exhibit 10); this implies that mix effects also are (or will be, given the lag between R&D spend and product flow) decelerating in absolute terms. Replacing the simplifying assumption of constant R&D productivity with the more realistic assumption that productivity (at least in biopharma) is declining[6], it becomes even more likely that mix-driven growth in real health spending is in a decelerating secular trend. In hospitals[7] (Exhibit 11), measured growth in mix is in fact decelerating; since 2000 rolling 10-year growth in mix[8] has fallen by half, from 4% to 2%. In-line with this, for our immediate purposes of estimating secular effects, we assume that the contribution of mix to system-wide real demand growth also falls by half, from an estimated 1.1% to 55bp

As an aside, it’s also useful to consider mix effects in relative terms. Medical research as a percent of total US research appears to have steadily growth before reaching a plateau around the year 2000 (Exhibit 12). This relative plateau is evident in two respects: medical R&D spending as a percent of total R&D spending; and, medical patents issued as a percentage of total patents issued. Growth in mix is one of several reasons health spending has grown faster than the broader economy over the last several decades, and medical R&D’s pre-2000 rising share of total spending and patents issued was presumably the main driver of mix. Now that healthcare no longer appears to be innovating more rapidly than the broader economy, mix growth is less likely to contribute to any excess growth in health spending relative to GDP. This implies that innovation has become a selective driver of real growth within healthcare, whereas in the past it was very nearly a categorical driver

Secular effects on real pricing power

Because most healthcare is purchased by the insured, and because most of this healthcare is purchased by persons who have exceeded their deductibles (Exhibit 9, again), the price elasticity of demand for healthcare is remarkably low – on the order of 0.17. Factoring in the relative absence of price controls and a relative predominance of profit-maximizing producers, even if we knew nothing about historical pricing trends we would naturally assume that US healthcare pricing consistently grows in real terms[9] — and this has plainly been the case (Exhibit 13)

We genuinely believe that real price growth in US healthcare is that simple: the combination of inelastic demand and profit maximizing[10] producers with limited or no price controls[11] cannot be expected to produce any other result. And, in the narrow context of secular effects we see nothing that is likely to slow the real pricing trend. Looking ahead to our analysis of reform effects in Part 4, we expect to show that reform actually raises the system’s inflationary gearing; more specifically, we believe that declining elasticity as a byproduct of more generally available coverage will have a larger upward effect on prices than can be countered by the reform Act’s limited attempts to reduce price growth

  1. Mathematically, mix effects are the excess cost of today’s standard of care for a given condition as compared to the cost of yesterday’s standard of care for that same condition. The difference in cost is of course driven by the difference in price between the old and new standard of care – so, to the extent the new standard is over-priced relative to its incremental value over the old standard, it can be fairly said that growth in mix ‘as-measured’ is heavily influenced by real pricing effects
  2. For simplicity we assume growth in the uninsured is a consequence of falling ESI participation. Since 1987 an additional 5 percent of the population is uninsured. The relative health and age-adjusted compensation of ESI vs. uninsured households is approximately 2.7 : 1; i.e. the uninsured consume about 63% less care than the insured. A 63% drop in consumption for 5% of households over 23 years is an average consumption headwind of 15bp annually
  3. Some very tightly managed networks, such as vertically integrated health networks, do offer lower premiums than average for comparable amounts of care; however these options are not available in most markets
  4. High deductible health plan
  5. We measured the real trend for PhRMA members, for constituents of various indices of innovative companies (pharma, bio, and others), and also for non-commercial (e.g. university, NIH, etc.) spending. Each time series supports the same conclusion, namely that real R&D spending has decelerated
  6. See “Below Zero and Falling Fast: R&D Productivity as an Enterprise-Wide Crisis,” SSR, October 3, 2011
  7. To calculate mix, we have to be able to back out unit demand; because of robust days/capita data across a considerable timespan we can do this more easily in hospitals than elsewhere. Also, because hospitals are effectively a cross-roads of care delivery, it’s reasonable to believe the hospital mix trend is in fact reflective of the broader health system’s mix trend
  8. Anticipating the obvious objection – that mix has decelerated alongside economic deceleration – we would emphasize that the primary effect of economic downturns on US healthcare demand is to decrease the intensity of demand, i.e. the number of units consumed per person. Our research suggests that mix effects are relatively independent of the underlying economic cycle
  9. A profit maximizing producer facing no price controls in an elasticity environment of 0.17 can expect to increase revenues 8.1 percent for each 10 percent increase in price (8.3% more volume at a 10% higher price)
  10. We recognize that quite a lot of care is delivered by non-profit entities, particularly hospitals. However, even non-profit hospitals tend to be ‘mission-maximizing’, i.e. we believe these entities generally tend to scale spending to available revenues in an attempt to deliver as much care as possible
  11. This is by no means a criticism of profit motive or an appeal for price controls – we’re simply trying to make clear that the combination of price inelasticity and profit motive reliably produces real price growth in the absence of controls
Print Friendly