Accelerating Growth in Hospitals’, Physicians’ Offices and Other Care Settings’ Labor Hours Signals Improving Healthcare Demand

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

Richard Evans / Scott Hinds / Ryan Baum

203.901.1631 /.1632 / .1627

richard@ / hinds@ /

twitter.jpg @SecSovHealth

March 12, 2012

Accelerating Growth in Hospitals, Physicians’ Offices and Other Care Settings’ Labor Hours Signals Improving Healthcare Demand

  • Reported earnings are a surprisingly weak indicator of near-term changes in healthcare demand for many reasons, the most important being that the dominant settings of care (hospitals, physicians’ offices) are vastly under-sampled
  • In contrast, monthly aggregate labor hours offer a timely and broad-based sample of activity in hospitals and physicians’ offices; and, because these settings do not produce inventory, labor hours (as a proxy for supply of care) are a useful coincident indicator of healthcare demand
  • In settings of care broadly, and in hospitals and physicians’ offices specifically, aggregate labor hours have recently begun growing at an accelerating rate, and this strongly suggests improving US healthcare demand
  • Volume-sensitive sub-sectors, in particular hospitals and non-Rx consumables, continue to imply a continuation of trailing demand levels; this despite the facts that 2010 was a 50-year low in per-capita demand, that these sub-sectors are substantially more volume-levered than their healthcare peers, and that healthcare demand appears to be accelerating
  • We continue to believe hospitals (e.g. HCA, UHS, THC, HMA, LPNT, CYH) are generally undervalued, as are non-Rx consumables (e.g. BAX, COV, BDX, BCR, CFN)
  • Despite evidence of rising utilization, we also continue to see the commercial HMOs (e.g. UNH, WLP, CI, AET) as undervalued. Because utilization gains correspond with employment gains, HMOs gain enrollment, and it is well established that marginal enrollees tend to be considerably healthier than legacy enrollees in an employment up-cycle. It follows that HMOs stand to see both enrollment gains and stable gross margins, where current valuations imply margin compression

Trailing measures of healthcare demand are weak for cyclical reasons

We believe the trailing-period slowdown in US healthcare demand is a predominantly cyclical phenomenon, which implies a rebound in healthcare demand as the economy improves. We readily accept that healthcare faces both secular and cyclical headwinds, and that both matter; however we’ve shown that in the present case cyclical headwinds appear dominant, having at least eight times the immediate influence of secular headwinds[1]

US healthcare demand growth can be broken into components which are plainly cyclical (per-capita changes in ‘units’ and ‘mix’ of care), and those which are not (medical inflation, innovation, aging, and population growth). By backing out the non-cyclical components of healthcare demand we’re left with an aggregate cyclical measure we term ‘per-capita elasticity / mix’, and changes in this measure appear highly correlated with the economic cycle (Exhibit 1). This cyclical measure reached historic lows in 2010, and the change in this measure (down 1.6%, from 1.2% to -40bp) from its trailing pre-recession (2000 – 2008) rate accounts for more than all of the 60bp fall in real demand versus the pre-recession baseline, and is very nearly as large as the 1.8% fall in real demand versus the 50 year average (Exhibit 2)

Aggregate healthcare labor hours signal an improving demand trend

Identifying a turning point in healthcare demand before it’s evident in share prices obviously requires data that are more timely than the 50+ year annual series underlying the prior 2 exhibits. Disappointingly, a roll-up of bellwether companies’ publicly reported quarterly sales and earnings is a surprisingly weak near-term indicator of changes in total US healthcare demand. Hospitals and physicians’ offices are poorly sampled: non-profit facilities dominate total US hospital beds, and few if any pure-play physician practice settings publicly report sales and earnings. Product and service suppliers’ results generally are too idiosyncratic to point to a clear trend; even distributors’ footprints are generally too narrow to quickly and accurately reflect system-wide changes in demand

Fortunately, hospitals’ and physicians’ offices use of labor is accurately[2] and frequently recorded. The Bureau of Labor Statistics’ (BLS) Current Employment Survey (CES) collects monthly data on employment, hours and wages at the establishment[3] level, and establishments are classified in such a manner that we can observe near-term use of labor in discrete types of healthcare settings. Crucially, the settings of care[4] measured by the BLS do not produce inventory; correspondingly, aggregate hours worked in these settings should be an accurate coincident indicator of both supply and demand

Exhibits 3, 4, and 5 show the annual rate of change in aggregate hours (green data series) worked for Healthcare settings broadly, General Medical and Surgical Hospitals, and Physicians’ Offices respectively. Data in each graph are current through December of 2011; in each case the current employment rate is shown for comparison. In all cases we see recently accelerating growth in aggregate work hours. In all but two instances we see reasonable correlation between changes in aggregate works hours and changes in the employment rate. The first exception: the latter half of the 90’s sees falling growth in aggregate hours despite rising employment; this corresponds to the timeframe in which HMO influence peaked, and we speculate that HMOs’ brief but significant dampening of total demand accounts for the late ‘90’s dip in hours worked. Far more narrowly, as a second exception we notice a rise and fall in aggregate hours in Physicians’ Offices after the beginning of the recession but before a more recent acceleration of growth in aggregate hours; at the risk of simply being fooled by noisy data, here we speculate that physicians’ efforts to make practices compliant with minimum electronic medical record (EMR) standards ahead of Medicare incentive deadlines may be the underlying cause

Investment Conclusion: Overweight volume-levered subs-sectors

The dominating effect of cyclical forces on immediate healthcare demand, and an apparent improvement in the demand trend co-incident with gradual but significant economic strengthening, contrasts with our read of valuations across healthcare; the more volume-sensitive subsectors (particularly non-Rx Consumables and Hospitals) continue to imply historically low levels of per-capita demand growth, despite some improvement in relative (to healthcare) valuations since the start of the year (Exhibits 6,7)

We continue to see Hospitals and non-Rx Consumables as undervalued; both sub-sectors are highly volume-levered, and carry valuations that reflect both modest demand growth and limited operating leverage, even in the face of strong evidence of cyclical demand improvement. We also favor HMOs, despite the tendency for HMOs to sell off on rising utilization. Key to our HMO thesis is that rising utilization occurs for cyclical reasons, which implies that utilization and employment rise in parallel. Because rising employment grows commercial enrollment, and because marginal enrollees tend to be substantially healthier than existing enrollees, the net effect on commercial risk books is likely to be a combination of expanding enrollment and stable gross margin. Recognizing the tendency of HMOs to fall on evidence of rising utilization, though at the risk of cutting our recommendations too finely, we would emphasize Hospitals and non-Rx Consumables ahead of what we believe will become a widely accepted expectation of higher utilization, and HMOs once a return to more normal utilization levels is fully priced in. An important consideration is the impact of the Supreme Court’s late June / early July ruling on the Affordable Care Act (ACA); believing that HMOs and Hospitals should react in opposite directions to most but not all likely Court rulings, and being unable to handicap the specifics of the Court’s ruling, we recommend holding both Hospitals and HMOs – but neither sub-sector without the other – around the expected date of the Court’s ruling

  1. US Healthcare Demand Slow for Cyclical (i.e. Temporary) Reasons …”, January 12, 2012, SSR
  2. CES data is benchmarked annually each March via a census of Unemployment Insurance reports collected from all employers by states quarterly (and which cover ~97 percent of non-farm workers). Between benchmarks, monthly dynamics are estimated via the CES instrument which asks a stratified sample of establishments about their worker counts, wages, hours, etc. Subsequent robust corrections are made for establishment closings and openings. In March 2011, 42,738 Education and Health Service establishments were in the active monthly CES sample – and the 5.8mm employees at these establishments represented a full 29 percent of the employment benchmark. Historically, revisions to CES estimates due to late monthly filings and the annual benchmarking process have been minimal. From 1979 through 2011, total nonfarm annual benchmark revisions have been between -0.7 and 0.7 percent
  3. Employer or, for an employer with multiple locations, a single location of an employer
  4. Hospitals, both General Med/Surg and ‘Other’; Offices of Physicians, Dentists, Optometrists, Chiropractors, Specialty Therapists and Other Health Practitioners; Outpatient Care Centers; Medical and Diagnostic Laboratories; and, Ambulance Services
Print Friendly