Why Health Reform Legislation Won’t Pass; More About Insurers’ 2010 Margins
November 17, 2009
Why Health Reform Legislation Won’t Pass; More About Insurers’ 2010 Margins
- We no longer expect Congress to pass impactful health reform legislation this year, or even in this political cycle.
- Voter attitudes are shifting away from both Democrats and health-reform; placing the considerable number of Dems from conservative states and districts in increasingly untenable positions.
- Substantial time should pass before a vote on final legislation; the Senate is unlikely to vote on a bill by year end, and a final vote on a conference bill is virtually impossible before late January. If trends in voter opinion continue, this is almost certainly too late.
- On top of this, efforts to keep abortion as a neutral issue have failed. 20 pro-life House Dems have put the issue ahead of health reform, refusing to support legislation that does not completely bar abortion coverage in plans that receive Federal funding.
- Pro-choice House Dems outnumber their pro-life House Democratic peers nearly 8 to 1. As any final legislation will certainly be well to the right of the House bill, this means House progressives may be asked to support a final bill whose healthcare provisions they find lacking in appeal, and to surrender ground on choice in the process. We bet at least a few refuse.
- Immigration is a less potent but still meaningful wild-card; 20 House votes rely on the Senate ultimately agreeing to loosen its language and let illegal immigrants purchase coverage on the exchanges with their own money, and to extend subsidies not only to citizens, but to anyone who is in the US legally. 15 Senate Democrats hail from states that Obama either lost, or carried with a 5 percent margin or less. These swing-votes hold inordinate power, and are much more conservative on immigration than their blue-state peers; we believe they may balk at these House provisions.
- The House bill’s largest source of funding is an incremental tax on the wealthy, which the aforementioned 15 Democratic Senators from red- or swing-states cannot support. The Senate Bill’s largest source of funding is taxes on relatively more expensive ‘Cadillac’ health plans, which the union-friendly House cannot support. In effect, each has settled on a plan that the other cannot pass.
More on Insurers’ MLRs:
- In our last call, we showed that fear-of-job-loss accounts for higher than expected claims cost in ’09, and that this effect is gone by ’10 (if not already), suggesting better MLRs for insurers in ’10.
- In this call, we back-test our fear-of-job-loss thesis against historic MLR patterns, and show that MLR patterns predicted by our job-loss assumptions match quite closely with history.
- Influenza, both in ‘traditional’ and H1N1 forms, is a potent source of MLR uncertainty. The severe and prolonged ’09 season plainly contributes to ’09 MLRs; the effect in ’10 should be much smaller.
- The ’09 flu season started early, and appears to be ending early, meaning flu-related claims pressures are likely to abate by 1Q/10. In the back of ’10, we expect greater availability of vaccine; in ’09 vaccine supplies only began to ramp as flu incidence began to fall. To date, 27.8 M H1N1 doses have shipped; this compares to 100 – 115 M at this point in a typical flu cycle.
- Insurers, particularly large-capitalization insurers with exposure to the employer-sponsored insurance markets, remain our most favored group, followed by PBMs.
We believe the odds of Congress passing impactful health legislation in this political cycle are increasingly remote, for four main reasons:
- voters’ attitude are shifting; as of November Republicans lead Democrats in the ‘generic’ 2010 Congressional ballot, and oppose health reform 48% to 43%;
- attempts to keep the abortion issue out of the debate have failed; the issue now exerts a polarizing influence, stands to weaken House support for the bill, and provides air-cover to Nelson (D-NE) for a nay vote in the Senate;
- immigration – because of a solid 20-vote Hispanic caucus the House cannot pass a bill that bars unregistered immigrants from purchasing coverage with their own money; this provides potential air-cover to as many as 5 nay votes from Senate Democrats who have conservative immigration records and hail from states that supported McCain in ’08; and,
- irreconcilable pay-fors – the Senate bill relies on taxes placed on high-cost health plans, which unions strongly oppose, and unions hold considerable sway in the House. The House bill relies on taxing the wealthy, which the Senate generally opposes.
Time – Shifting Voter Attitudes Likely to Pull Susceptible Members Out of the Yea Column
As voter attitudes shift away from Democrats, and into net opposition of health reform legislation, we believe that this exerts considerable pull on members whose ideologies conflict with provisions of the legislation, and/or whose constituents are likely to oppose the legislation.
85 percent of House Democrats voted ‘yea’ on H.R. 3962 (Exhibit 1); we sort these yea votes according to ideology and local politics. Blue Dogs, other centrists, and Pro-Life Democrats were less likely to support the bill than their Democratic peers (also Exhibit 1); we term these groups’ 49 yea votes as ‘ideologically vulnerable.’ Democrats in districts where McCain either carried the ’08 vote or lost by less than 10 percent were less likely to vote yea than their peers in districts that went more convincingly for Obama (again, Exhibit 1); we term these 39 yea votes ‘politically vulnerable.’ The intersection of ‘ideologically’ and ‘politically’ vulnerable contains 24 yea votes. Keeping in mind that H.R. 3962 passed by 2 votes, 24 vulnerable votes is quite a large number. As an aside and contrary to conventional wisdom, members’ tenure seems to have little effect; these ideological / political influences among freshmen and sophomores mirror the picture in the broader Congress (Exhibit 2).
In the Senate, we focus our analysis on Democrats hailing from States that either supported McCain in ’08, or who supported Obama by a narrow margin. 17 Senators fit this definition, though only 3 of these are up for re-election in 2010 (Exhibit 3).
Turning to voters, the New England Journal of Medicine recently published a comparison of voter attitudes in the ’94 and ’09 health reform debates (Exhibit 4). The analysis compares April of ’94 – which in retrospect was plainly a tipping point from support to opposition – to voter attitudes from October of this year. The similarities are remarkable.
More to the point, we find that the momentum in voter attitudes appears to be working against passage of reform. By extension, assuming these trends continue, the longer they have to play out, the more susceptible members’ votes are lost, and the less likely reform legislation is to pass.
Remarkably, voters’ belief that it is the Federal government’s responsibility to ensure that Americans have health insurance has shifted from a strong ‘yes’ to a weak ‘no’ (Exhibit 5). Who voters feel should be responsible for health insurance we do not know, but the apparent fact that Federal government is no longer the contemporary answer says much about the prospects of reform legislation passing.
Similarly, voters’ support for Congress’ immediate legislative efforts has eroded considerably, even since October of this year (Exhibit 6). On net, voters now oppose the passage of health reform legislation. And, attitudes on health reform legislation are closely linked to party affiliation; consistent with the recent shift away from support for health reform, voters now narrowly favor Republicans over Democrats for the ‘generic’ 2010 Congressional ballot (Exhibit 7), and much of this appears due to Republican gains among independents. This general trend of support for health reform legislation eroding with time is repeated across a broad variety of polls (Exhibit 8).
Keep in mind that poll data tend to be analyzed and reported using registered voters as the denominator, but that the make-up of actual voters shifts between Presidential and mid-term elections. Seniors are more highly represented in mid-term elections (Exhibit 9a); tend to be Democrats’ weakest demographic group (Exhibit 9b), and tend to oppose reforms (just 27% of seniors said they would be better off if health reform passed, according to the October 2009 Kaiser Health Tracking Poll). All of which argues that the pro-Republican / anti-health reform trends seen in poll data are likely to be even more apparent in the 2010 election. This tendency of seniors to be over-represented in the mid-terms is well known, and so carries weight with members facing the mid-term ballot in 2010.
This leads to the matter of timing: how long will the debate persist, and in particular will it last long enough for shifting attitudes to force ideologically and/or politically susceptible Democrats to vote no? The
legislative ball is in the Senate’s court, making the Senate’s calendar an appropriate backdrop to the timing question. The Senate has blocked off the week of November 23 for Thanksgiving; assuming the Senate convenes every weekday until Christmas – including Christmas Eve – that leaves 18 legislative days until Christmas. As is traditional, we’re assuming members are home at some point over the Holidays; in past years the pattern has been to adjourn in the 2nd or 3rd week of December, and return in the first week of January. To our minds that trip home is the critical event; putting members back in direct touch with their constituents arguably serves to drive home constituents’ views all the more clearly. More crudely, if members are back in their districts / states before a final vote on a conference bill, we’d expect more of the at-risk votes in either chamber to shift into the no column.
And there is practically no way to get to a conference vote by year-end. At best, the Senate might vote out its version of the legislation, but even this seems quite remote. Even a weak (sub-40 vote) Senate minority can create procedural hurdles that are ‘worth’ roughly 15 calendar days (Exhibit 10). And, assuming Senator Lieberman’s (I-CT) threat to filibuster a bill with a public option is made in earnest, Senate opponents of reform legislation appear to have the power to extend debate quite a bit further.
As a potentially important aside, even if Republicans were to mount a weak filibuster, consuming roughly 15 legislative days, this uses most (and arguably all, assuming any Holiday break whatsoever) of the Senate’s remaining 2009 calendar. If the majority invokes cloture in late December, it’s important to recognize that a maximum of 30 in-session hours remain to consider that specific legislation once cloture is invoked. Each Senator has the right to be recognized even after cloture, though the process of being recognized and the subject matter one can speak on is considerably narrowed. Nevertheless, though the Senate rarely consumes the 30 hours, a mini ‘post-cloture’ filibuster could run out this 30 hour clock, potentially adding several days to the more easily achievable 15 days of classic procedural delays. Assuming the Republican’s tactical objective is to have the Senate adjourn for the Holiday without a healthcare vote – and we believe it is — very late in December these marginal procedural delays could be meaningful.
Further on the matter of time, we note Senate Majority Leader Reid’s argument that CBO scoring is a considerable source of delay – though we’re not ready to accept it as a valid reflection of what is happening in the Senate. CBO produced a final estimate of the House bill on November 6th, within 9 days of the bill’s introduction in the House on October 29th. Similarly, CBO produced a detailed preliminary scoring of the Senate Finance Committee’s amended bill on October 7th, within 7 days of the bill having grown by 101 amendments during mark-up. Moreover, we’d note that CBO’s scores have been offered within uncharacteristically narrow economic constraints – both the current and immediate past Directors have been thoughtful and vocal on the big picture consequences of health costs in the past, but silent on these matters in the context of CBO scores. Accordingly we’d argue the CBO, if anything, is accommodative toward health reform legislation.
If we frame CBO’s task as scoring the merger of the Finance Committee mark-up with the Senate HELP bill – which passed committee on July 15th – then there is no reason to believe that a compliant CBO needs a period of weeks to produce a score. The far more likely explanation is that the Senate simply has not given CBO a final bill to score, which in turn suggests that the Senate has yet to find a pathway to 60 votes – all of which further argues that the reform debate lasts through year end.
Assuming opinions continue to trend away from Democrats and health reform, we see more than enough time for marginal Democratic votes to be lost to both ideological and local political pressures, and so view the interplay between timing, shifting voter attitudes, and the tenuous ideological and political positions of many Democrats as the most powerful reason that reform efforts may fail.
On Top of Everything Else, This is Now an Abortion Fight ….
A group of pro-life Democrats, lead by representative Stupak (D-MI), successfully negotiated for inclusion of language that bars the use of health insurance subsidies for abortion. The amendment effectively ensures that any public option, or any plan offered through a health insurance exchange (HIE), cannot offer coverage for abortion. The amendment was added to H.R. 3962 late on the Saturday before the bill went to the floor for voting, giving pro-choice opponents little opportunity to react. The Stupak language contrasts with more permissive language in the Senate Finance bill (Exhibit 11).
The core of Stupak’s group appears to consist of 20 pro-life Democrats (Exhibit 12). By comparison, House Democrats are decidedly pro-choice — 178 House Democrats voted for H.R. 3962 but against the Stupak amendment; and, these Democrats’ carry an average Planned Parenthood (pro-choice) score of 98.9, out of a possible 100. Pro-choice reaction to the Stupak amendment was decidedly negative, thus it is quite possible that if the Stupak language remains as is, that some pro-choice Democrats’ votes may be lost on a conference bill if pro-choice pressures build. Moreover, assuming pro-choice advocates were to push back against the Stupak language, the issue may give air-cover to a potential no vote from Senator Nelson (D-NE, Exhibit 13). Note that Senator Nelson has been vocally ambivalent about the reform bill, that McCain won Nebraska by a margin of 15 points; and, that Senator Nelson has a pro-life voting record (Exhibit 13).
Once out of the bottle in the context of health reform, abortion is an issue that’s very difficult to place back into a neutral role. The pro-life argument is essentially that any expanded Federal expenditure for health coverage will result in Federal dollars being spent for abortion unless recipients of subsidies are specifically barred from using these dollars for abortion. Stated differently, pro-life argues that no accounting segregation of federal and non-federal dollars can work, since the federal subsidies free non-federal dollars to be used for abortion, which in effect is a federal subsidy. The pro-choice argument is that barring use of subsidies for abortion represents a loss of choice, both for women whose non-Federal coverage is replaced by Federal coverage, and for women whose existing insurance carriers cease covering abortion in order to be eligible to offer Federally subsidized healthcare coverage. Both sides are right, but both sides cannot win – either there is a ban, or there is not; this is a perfect zero-sum game. If enough pro-choice Congressional Democrats prioritize choice over health reform – as Stupak’s bloc has prioritized its pro-life objective, then health reform cannot pass the House.
After it is merged with anything the Senate may pass, the liberal / progressive appeal of the original House bill’s healthcare provisions will surely be watered down. Thus at a minimum, we expect liberal / progressive House Democrats to be forced into a choice between accepting a much less appealing healthcare bill that also sacrifices ground on right-to-choose, or rejecting the conference bill entirely.
… and an Immigration Fight as Well
The 20-member Congressional Hispanic Caucus successfully negotiated for an easing of restrictions on illegal immigrants in H.R. 3962, having threatened to vote against the legislation if their concerns were not addressed. As a result, H.R. 3962 now allows anyone to purchase health coverage through health insurance exchanges (regardless of immigration status), and affordable credit subsidies are made available to US citizens, US nationals, and individuals that are lawfully present (Exhibit 14, including comparison to Senate language).
In contrast to the abortion issue, we see little pressure from House Democrats to roll back this language; rather, we see the immigration issue potentially having its effect in the Senate. Returning again to our list of Democratic Senators in Republican states or states carried by Obama on a narrow margin, we find 5 that carry relatively high (i.e. anti-immigration) scores from the Federation for American Immigration Reform (FAIR, Exhibit 15). Senator Landrieu (D-LA) in particular is a prominent fence-sitter on heath reform; arguably the more immigration-friendly language in the House bill, particularly if this language survived into a conference bill, would provide considerable air-cover for a no vote. Conversely, if the Hispanic caucus’ language is not in the conference bill, the group’s 20 votes are quite likely to be lost.
Union influences among House Democrats are considerable; those voting yea on H.R. 3962 have an average AFL-CIO score of 98.8 out of a possible 100. Here, the problem is that the Senate bill relies heavily on taxes applied to high-premium or so-called ‘Cadillac’ plans. Unions strongly oppose this source of pay-fors; having sacrificed wages for health coverage, unions’ health benefits are comparatively generous, and accordingly may be disproportionately at risk of falling subject to such taxes.
Conversely, the House bill relies heavily on taxing the wealthy. Here, the problem is the political position of moderate Democratic Senators that for all intents and purposes control the fate of healthcare legislation in the Senate. As we showed in Exhibit 3, 12 Democratic Senators are from states that McCain won; 15 are from states that either McCain won, or that President Obama carried by less than 5 percent. The prospect of a new tax in general, and a tax that is focused on the wealthy in particular, is a Republican anathema. Accordingly, we believe it’s unlikely that the Senate would pass a bill using the House’s pay-fors.
All of which suggests that each chamber has passed or is trying to pass a bill that the other chamber simply cannot.
Summary on the Prospects of Reform Legislation Passing
To our minds the preceding four arguments are independent of one another, and each represents a separate means by which reform efforts might be derailed. Taken together, we find the argument that reform legislation fails is now far more compelling than any argument that legislation passes.
We’re wholly aware of the political stakes for Congress and the Administration, and that these stakes only serve to increase the pressure to pass reforms. That said, constituents are the ultimate source of pressure, and constituents are increasingly in opposition to the reform effort. This is true not simply on Democrat v. Republican or liberal v. conservative lines; the introduction of abortion, immigration – and, effectively in the matter of pay-fors, class – as issues linked to healthcare reform means the initiative has even more ways to fail, and fewer to succeed.
We also acknowledge the prospect of an 11th hour bill being delivered by the White House to the Congress. It would have to happen before the Holiday break – i.e. now – otherwise trends in voter sentiment mean there may be insufficient Democratic votes available from traditionally Republican geographies. It would clearly have to neutralize the abortion issue, but we don’t believe that, in the context of Federal expansion of healthcare subsidies, that abortion is anything but a zero-sum game – i.e. there is no neutral ground. It would have to avoid stepping on the immigration issue – this may be possible, we see the issue of immigration less as a catalyst for failure, more as a convenient excuse for Democrats from more conservative geographies to oppose legislation that they don’t like for other reasons. Finally, it would have to pay for itself – and in so doing draw the majority of its funding from sources that are not yet in play in either the House or Senate versions. It’s not entirely unlikely that industry (particularly the healthcare industry) would be asked to pay even more, though this raises the likelihood that the industry’s already fragile support would turn to outright opposition. In short, we don’t think health reform is failing because someone hasn’t written the right bill; health reform is failing because no one created a durable coalition in the first place, and potential members of such a coalition have now been drawn into other (abortion, immigration, class) battles. We’re convinced that, at least for the time being, it’s simply over.
More About Medical Loss Ratios
In our previous call (Nov. 2, “2010 HMO and PBM GM’s Expand; The Public Plan Has the Blues”) we showed that greater-than-expected ’09 claims cost are likely related to job losses. In particular, we addressed the relative impacts of COBRA enrollment and fear-of-job-loss on MLR pressures in 2009 – 2010. In that call, we used MEPS data to demonstrate that across the ’01 – ’06 period (a) COBRA claims costs per enrollee were 63% greater than the claims costs of persons who remained employed; (b) persons losing their jobs had health claims that were 41% higher than persons who kept their jobs; and (c) the effect of fear-of-job loss on MLR’s is (potentially much) greater than the COBRA effect during periods of significant job loss, since the number of persons losing their jobs (and not enrolling in COBRA) is much greater than the number who enroll in COBRA. We concluded that fear-of-job loss does much to explain greater-than-expected claims cost trends; and, that these claims pressures should abate as job losses stabilize.
Three natural questions arise from these results: (1) do the implied results track with actual observed MLR results in the large capitalization insurance segment; (2) is there a trend among enrollees new to employer sponsored insurance (ESI) plans to over-consume, thus creating an offsetting ‘re-enrollment’ penalty; and (3) which categories of medical expenditure see the largest relative changes in consumption among COBRA and fear-of-job-loss enrollees?
Projected MLR Dynamics Match Observed Results
Below we are republishing our projected MLR fear-of-job-loss and COBRA effects (Exhibit 16, originally published in our Nov. 2 call). Note that in that previous call we concluded that job loss fears are at least twice as important as COBRA; and, that because such a larger proportion of the workforce has lost jobs, that the total degree of fear-of-job-loss MLR pressure is much greater now than in ’01 – ’03 (where we very roughly estimate an effect of 3.2%).
In order to validate this estimate, we have combined our relative consumption estimates with quarterly U.S. job loss data is order to generate implied MLR dynamics during the 2001 – 2009E period. Comparing the resulting forecast to actual segment MLRs over the same period, we find that our job loss model tracks with the sector extremely closely, particularly since 2004. The relationship between predicted and actual MLR in 2002 – 2003 is directionally indicative, and the relationship in 2004 – 2009E is very strong (Exhibit 17). Over this later period, the correlation coefficient between the predicted and actual annual changes in MLR is 0.87 (R-square of 0.76).
We believe that these strong correlations are further evidence of the role that fear-of-job-loss plays in insurers’ MLRs across economic inflections; and, by extension, have that much greater confidence that MLR trends should improve in 2010.
No Re-enrollment MLR Penalty
We find no evidence that new enrollees make greater health claims than persons who keep their employer coverage (Exhibit 18). To the contrary, over the ’01 – ’06 period, the medical claims of persons who began the year uninsured but who obtained ESI coverage were 25% lower than those who kept their insurance. This result also reinforces our conclusion that insurers should see an alleviation of ’08 / ’09 MLR pressures as ’10 unfolds – even as workers who lost their jobs (and over-consumed on their way out, or on COBRA) find their way back into the workforce, we need not anticipate extraordinary medical claims as they return to ESI.
Over-Consumption Patterns within Lost ESI Populations
In order to track which medical claims are subject to the largest change among COBRA and fear-of-job-loss enrollees, we break total medical claims into nine mutually exclusive and exhaustive categories. The results, depicted in Exhibit 19, illustrate with more granularity the points of MLR pain for large insurers; or, conversely, the demand inflections experienced by providers.
Not surprisingly given the well-understood adverse selection problem of COBRA enrollment (i.e., they are sicker and more likely to consume healthcare services than the broader ESI population), those enrollees had the biggest disparity in medical claims for prescription drugs, office visits (both physicians, and non-physicians like PA/NPs, chiropractors, etc.), and hospital services (including both inpatient and outpatient procedures). Over the ’01 – ’06 period COBRA enrollees actually under-consumed ER services, relative to the employed group.
The fear-of-job-loss group had consumption nearly identical to employed enrollees in six of the nine categories, with the entirety of their overconsumption attributable to claims related to office visits, ER visits and inpatient procedures. Our hypothesis is that fear-of-job-loss consumption should be both discretionary and acute – the incremental consumption should be something over which the patient has a relatively high degree of control, and the commitment to consume should be short enough that spending is unlikely to extend into the post-coverage period. The consumption pattern is consistent with this thesis, but by no means can we argue that it is compelling support for our thesis. The MEPS dataset does not offer greater detail, so we can’t tell exactly what types of visits and procedures saw accelerated consumption, unless we can find other means of doing so. As an aside, while the tendency to over-consume ER visits might seem counter to our thesis, we’d note the tendency of younger, less well insured and lower-income socio-economic groups to use the ER as their source of primary care.
Flu and MLR’s
The flu season is a potent wildcard in claims cost, never more so than at present. 2009 has been both a severe and a prolonged flu season (Exhibit 20), thus the question becomes whether the 2009 season extends substantially into 2010, and whether the late 2010 season is anywhere near as severe as 2009. The happy news on the first count is that the percent of tests positive for flu are falling; this suggests that the 2009/2010 season may have been effectively left-shifted such that the majority of its claims cost fall in 2009 (Exhibit 21). As regards the question of relative severity in 2010, at a minimum we can argue that H1N1 vaccine availability is likely greater in 2010 than in 2009. Note Exhibit 22, which shows that vaccine availability is ramping up just as the percent of flu tests that are positive is ramping down – in effect the bulk of available H1N1 vaccine may be delivered after the season has largely run its course. Note also the cumulative doses shipped — as of last week 27.8 M doses had been shipped in the US; this compares to typical distribution of ‘traditional’ flu vaccine, wherein we would expect to see roughly 100 – 115 M doses of vaccine readily available by this point in the fall-winter season. Taken together, it appears as though the 2009 flu season has been unusually long, unusually severe, and unusually early, all against the backdrop of an unusually susceptible population. This argues that 2010 flu-related claims cost effects should be lower than those in 2009 – both at the beginning of the year (early 2009 season) and at the back of the year (2010 vaccine availability).