The Apparent Link Between Employment and Healthcare Demand

eytan
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

richard@ / hinds@ / baum@sector-sovereign.com

twitter.jpg @SecSovHealth

May 31, 2012

The Apparent Link Between Employment and Healthcare Demand

  • Privately insured US persons (and presumably households) consume 2.7x the healthcare of age- and health-status adjusted persons (and households) who are uninsured
  • As employment rises and falls, households shift from one category (un-/under-employed, un-/under-insured) to another (employed / insured), and this appears to be by far the largest driver of national changes in ‘unit’ demand for healthcare across the current cycle
  • Since 2008, job losses explain a 3.1% drop in per-capita unit demand among civilian working-age households, and at least a 2.2% drop in per-capita unit demand at the national level – i.e. very nearly the entire drop in demand
  • As employment rises the effect reverses; we estimate a 70bp increase in per-capita unit demand among the civilian working-age (and their beneficiaries) for each 1% gain in adjusted (for labor force declines) employment
  • This further underscores our thesis that healthcare ‘unit’ demand grows as the economy (and employment) improve; and, our related thesis that commercial HMOs are a value despite rising utilization (commercial HMOs benefit from rising enrollment, but are not exposed to the major source of rising utilization, thus gross margins can remain stable)
  • Since peak unemployment (October 2009) consensus forward (FY+2) earnings estimates for non-financial / non-health companies have grown steadily (as have share prices), consistent with improving economic conditions. In contrast, forward estimates for volume-sensitive healthcare names are relatively stable (as are forward PEs); this implies: 1) that the market is broadly willing to price in cyclical improvements in longer-term earnings power; and 2) that more narrowly, the market does not expect cyclically driven gains in earnings for the volume-sensitive healthcare names. In effect this argues that the market views weak trailing healthcare demand as a secular new reality, rather than what we believe it is far more likely to have been – a cyclical trough
  • We continue to favor a pro-US / pro-cyclical tilt to healthcare portfolios; in particular we favor Hospitals and HMOs (held together to hedge SCOTUS risks), and non-Rx consumables (e.g. COV, BAX, BCR, BDX, HSP, OMI, CFN)
  • Until recently we stayed away from Dental, Drug Distribution, Drug Retail, and Diagnostic Laboratories due to a variety of middle to longer term structural concerns (except for Dental where our concern had been purely valuation). Because we now feel better able to time the cyclical demand recovery – and in fact believe that it is ongoing – we see these sub-sectors benefitting from volume gains (and in the case of Dental, mix gains) before our structural concerns are likely to weigh on valuations

What job losses mean for US healthcare demand

US persons (< 65 y.o.) with private health insurance consume 2.7x the healthcare ‘units’ as compared to age- and health-status matched persons who are uninsured (Exhibit 1). This relationship appears to be independent of the economic cycle, and consistent across care settings with the exception of outpatient and emergency room (ER) visits (Exhibit 1, again). The outpatient and ER ratios are outliers simply because the uninsured are much more likely to seek care in ERs than in outpatient settings (Exhibit 2)

This 2.7x relative rate of healthcare consumption is a reasonable proxy for what happens when a household shifts from unemployed / uninsured to employed and insured – i.e. on an age- and health-status matched basis, we believe that shifting a person from uninsured (and presumably also un- or under-employed) to employed and insured produces an average 2.7x increase in that person’s ‘unit’ consumption of healthcare[1]

Knowing this, we can estimate the impact of lost employment on unit demand for healthcare. From 1996 through 2008, the civilian labor force was on average 66.5% of the population, and the unemployment rate was on average 5.0%. The ‘unadjusted’ (stated) annual unemployment rate peaked in 2010 at 9.6%; however if we adjust for the change in the civilian labor force we estimate an ‘adjusted’ unemployment rate of 11.4% in 2010, as compared to an adjusted unemployment rate of 6.3% in 2008. Applying the relative rate of 2.7x consumption for privately insured v. uninsured, we estimate that job losses since 2008 reduced per-capita unit demand (among working age civilian households) for healthcare by roughly 3 percent (Exhibit 3). As a rule of thumb, a 1 percent gain in adjusted employment translates to a 70bp gain in average per-capita unit demand for healthcare among working age civilian households

These households account for about two-thirds of total healthcare demand. Therefore even if no other demographic groups were demand-elastic, we would expect the peak 3.3% reduction in per-capita unit demand among the civilian working age to result in a roughly 2.2% drop in national average per-capita unit demand as compared to the pre-2008 period. This is very nearly large enough to explain the 2.4% drop in per-capita ‘elasticity and mix’ seen in 2009 – 2011 versus the preceding 50-year average (Exhibit 4). In reality we know that even per-capita Medicare demand is elastic, i.e. cyclical, and that the per-capita ‘elasticity and mix’ measure under-estimates drops in ‘pure’ per-capita unit demand[2]; so the ‘true’ national drop in unit demand attributable to the economic slowdown should lie somewhere between these 2.2% and 3.3% values

Why (national average) healthcare utilization can rise without negatively affecting HMOs

Based on the preceding, we believe shifting households into and out of employment and employer-sponsored health insurance explains the overwhelming majority of changes to per-capita healthcare demand since 2008. As a household gains employment and insurance, its healthcare consumption on average nearly triples; accordingly national average per-capita demand will improve as job growth continues. Crucially, this primary driver of rising national average utilization does not affect commercial HMOs, who are exposed only to those (presumably much smaller) utilization gains that may occur in households who already had employment and insurance (and whose confidence inevitably is rising as the economy improves). And, we believe the two benefits to HMOs of rising employment – enrollment growth drives both revenue growth and an improving risk profile of the marginal enrollee[3] – more than outweigh any tendency of members’ healthcare consumption to rise slightly with improving consumer confidence

Are volume-sensitive healthcare names weak because the market is broadly skeptical of an economic recovery, or because the market is more narrowly skeptical that healthcare demand will improve?

We compared forward (FY+2) EPS estimates for volume-sensitive healthcare names to all US-listed non-financial / non-healthcare names with market caps > $1B, using October of 2009 (trough employment) as a starting point (Exhibit 5). Consistent with improving economic conditions, estimates for non-financial / non-health names have risen substantially since the apparent trough of the recession; however forward estimates for volume-sensitive healthcare names have remained far more conservative. Bearing in mind that nominal FY+2 estimates naturally tend to rise in absolute terms at year-end as another calendar year comes into view, very little of the forward estimate growth for volume-sensitive healthcare names appears to have come from actual upward revisions, particularly as compared to the positive revisions in the broader basket of non-financial / non-health names. Absolute and relative fPEs on FY+2 consensus have been reasonably stable for both groups (Exhibit 6); this implies that the market has priced in the rising expectations for the non-financial / non-health names, and has similarly shared in the sell-side’s skepticism regarding any substantial recovery in healthcare unit demand

Taken together, these findings suggest that the broader market is willing to price in improving expectations, i.e. that weakness in volume-sensitive names has more to do with healthcare-specific concerns than with concerns about the broader economy. In particular we believe estimates and share prices for the volume-sensitive healthcare names reflect a consensus belief that trailing unit demand has been weak more for secular than cyclical reasons, and that utilization will not improve with rising employment. Plainly we don’t share this view, and so continue to believe that a pro-US / pro-cyclical tilt to healthcare portfolios is called for

  1. ‘Person’ here refers to either the actual employee or a member of a household in which the head of household has gained employment and insurance where none was had before
  2. ‘Per-capita elasticity and mix’ is a measure we back into using National Health Expenditures (NHE) data by removing the effects of real pricing gains, population growth, and changes in average population age. The NHE data are our best source of demand data across multiple economic cycles, but offer no way of separating mix from the per-capita unit measure. Because mix and real price tend to grow largely without regard to the underlying economic cycle, we suspect the 2.4% drop in the ‘per-capita elasticity and mix’ measure (2009 – 2011 v. 50-year average) understates the true drop in ‘pure’ per-capita unit demand. Separately, note that the earlier per-capita elasticity / mix low of +30bps in 1990 – 1999 corresponds to the timeframe during which hospital lengths of stay were falling rapidly because of changes to reimbursement
  3. New employees generally consume less care than ‘legacy’ employees, but pay the same premiums. We recognize that it sounds as though we’re trying to have it both ways – we’re arguing new employees consume more than the uninsured thus raising national average utilization, but less than ‘legacy’ employees thus improving commercial HMOs’ marginal risks – but the numbers bear us out
Print Friendly