Blog Home


Cadillacs Or Ambulances? The Senate Tax On ‘Excessive’ Benefits

December 3rd, 2009
by Joseph White and Timothy Jost

Editor’s Note: For more on the controversy over taxing high-cost health plans, see “Taxing Cadillac Health Plans May Produce Chevy Results,” a Health Affairs Web First article by Jon Gabel and coauthors published today.

The Senate “Patient Protection and Affordable Care Act” draft legislation includes a steep excise tax on high-cost, so-called Cadillac insurance plans.  This provision is strongly supported by many economists and policy commentators, who believe that the tax could make the financing of U.S. health care more equitable and its provision more efficient.  Yet both propositions are highly questionable.

Mistaking Cadillac Prices for Cadillac Plans

The intent of the bill is to limit ability to pay for health insurance with pre-tax dollars.  The tax covers both employer and employee payments for health care, whether as premiums or self-insured expenses, including not just “health insurance” but dental, vision, flexible spending account, and any other such benefits.  In 2013, if pre-tax spending exceeds $8,500 per year for an individual or $23,000 for a family, the value of coverage above the threshold would be taxed at a rate of 40%.  For example, if an employer self-insured, and its average expenses for employees who had family coverage were $25,000, it would pay an $800 penalty for each covered employee.  Beginning in 2014, the thresholds would be raised each year by the annual increase in the Consumer Price Index (CPI) plus one percentage point.

How high will these thresholds be compared to average premiums?  That depends on the difference between health care cost increases and the CPI.  We assumed that the CPI would maintain its trend of 2.5% growth per year; used Kaiser/HRET data for a 2009 premium baseline, and assumed that premiums would grow at the 6% per year average of 2004-2009. We then could expect the threshold to be just less than 40% higher than average premiums in 2013, and about 20% higher in 2019.  The average family premium would be $16,685 in 2013 compared to a penalty threshold of $23,000; and $23,952 in 2019 compared to a threshold of $28,273. 

CBO and advocates for the excise tax assume that premiums will grow less dramatically, but there is no evident basis for that assumption.  In fact, premiums appear to be growing much more rapidly this year.  Also, the Kaiser/HRET data understate coverage costs because they do not include separate dental or vision plans.  If we assume that the expenses for one author’s fairly standard Anthem PPO and Dentemax PPO coverage through Case Western Reserve University were to grow at the 2004-09 rate after 2010, that coverage would become “excessive” in 2019.  Does that mean that the benefits are “too rich” now?  If not now, when?  And why then?

In short, the bill defines a Cadillac price, not a Cadillac plan.  Just as the Alternative Minimum Tax started as a tax on the very rich and eventually extended to much of the middle class, the “Cadillac plan” tax will over time cover an increasing number of mid-range sedans as well.

Confusing Cadillacs and Ambulances

Why, in fact, do some plans cost more than others now?  As Paul Van de Water reports, some offer especially generous benefits, as with coverage for top executives of Goldman Sachs.  But coverage could be expensive because the people covered are unusually sick (on average) – because the vehicle is not a Cadillac with “luxury” or “unnecessary” features, but an ambulance.  A plan could have a typical benefit package but disproportionately older or riskier members, or employees within high-risk occupations, or be in a market where medical costs are particularly high through no fault of those workers. 

Premiums vary significantly across markets, by age of enrollees and by type of employment, and insurance companies vary their rates accordingly.  They have convinced the Senate drafters that it is only fair to allow premiums to vary by a factor of three according to age – far more than the standard for “excess” costs compared to average costs, even in 2013.  The Senate excise tax plan addresses such concerns in an entirely inadequate way.  It allows higher thresholds for three years in the 17 most expensive states – but markets are not at the state level.  It allows higher thresholds for occupational groupings that are far broader than insurers’ rating bands (and in a way that likely favors rural states, not surprisingly).  It does nothing about age.  Therefore benefits deemed “excessive” due to their cost for many groups are likely to offer average protection from medical risk.

How serious could this inequity be?  We do not know – but neither do advocates for the excise tax, who, when asked, agree that there are no good data on which plans cost how much and for what reasons.  We do know that variations due to risk profiles of groups often exceed the difference between average costs and the thresholds, and will do so increasingly over time.

Claims about Savings

The excise tax should have a positive effect on the federal budget balance.  If the amount paid for any group of employees is deemed to exceed the targets, either the federal government will receive the excise tax, or the employer will reduce the benefits.  If money not paid in benefits is paid either as wages to employees or profits to investors, it will be subject to tax.  Regardless, the Treasury gains.

We do not know, however, what the distributional effect of those new revenues will be.  It depends on why a given group has “excess” costs.  If the goal is to make medical care financing more progressive then a targeted tax on high incomes, such as is in the House bill, would have far more certain effects. 

Many advocates for the excise tax argue, however, that the tax will drive down health care costs overall.  CBO Director Douglas Elmendorf calls the tax a powerful policy lever to reduce costs; Paul Ginsburg argues that it “is likely to have a powerful effect on health care costs by inducing people to shift to less-comprehensive insurance.”  Other things being equal, less insurance should mean less spending, because some people will not be able to afford as much care.  The question is whether that is the best or even an advisable way to save money.  (Moreover, other things are not equal.  That is why the many countries with more extensive insurance than the United States nevertheless have lower spending.) 

If the “excess” being trimmed consists of unusual benefits such as health club memberships, then trimming that fat will have no effect on the costs of more standard care.  So the argument about excess should focus on the effect of higher cost sharing.  Advocates believe that this will reduce costs painlessly, because people will reduce current consumption that has no net benefit.  The empirical basis for this assumption, however, is surprisingly weak.

Advocates usually point to the RAND Health Insurance Experiment (HIE).  The HIE found that people who had to pay more out of pocket for health care used less care.  Moreover, the average person who had higher cost sharing and used less care did not suffer measurable ill effects.  This finding has had a deep and lasting effect on the economics profession.  Yet the effect appears to have exceeded the evidence.

Those who used less care did so pretty much across the board, forgoing care both “necessary” and “unnecessary” in the researchers’ judgment.  This raises questions about whether the forgone care might have led to worse results if the study had lasted longer.  Moreover, cost reductions largely stemmed from people not initiating care, rather than from their using fewer services once care had begun.  The argument for the excise tax based on the HIE, then, should be that with higher cost sharing, people will forgo going to the doctor, and that will work out overall, an argument rarely made explicitly.

Forgone care did have worse results for low-income enrollees.  Moreover, the rate of attrition during the study from the cost-sharing arm was 16 times greater than that of the free-care arm, suggesting that some who needed care may have reverted to their pre-experiment coverage to avoid paying the cost sharing.  The study authors argued that the demographic characteristics of the people who dropped out were similar to those of the people who stayed in the study; but the evidence about risk adjustment establishes that demographic variables are not a sufficient control.  So even the conclusion that higher cost sharing was not related to worse outcomes for the non-poor could be in error.

There is good reason to question whether the HIE even provides data that should be used to predict the effects of systemwide reforms.  There were too few enrollees in the HIE for their lesser demand to provoke any response from providers wishing to maintain incomes.  Higher cost sharing systemwide could lead to hospitals, for example, raising prices for nondiscretionary care, thereby eliminating much of the savings from cost sharing.  (Jeff Richardson, “The Effects of Consumer Co-payments in Medical Care,” Background Paper 5 (Canberra, Australia: National Health Strategy, June 1991))

Moreover, as RAND researchers summarized in a 2006 overview, the health insurance system has changed significantly in the thirty years since the HIE began, so it is impossible to know whether a similar experiment would have the same results today.  A series of studies have identified situations in which lesser cost sharing improved health or even saved money, compared to greater cost sharing.  One suggested that the chronically ill, especially, should have lower cost sharing for some services.  But that implies that any penalty for “excessive” costs should depend on the health status of the group covered – and the Senate proposal includes nothing of the sort.

Another argument claims that the excise tax would cause employers to fight harder to obtain lower prices for coverage, giving insurers more reason to try to restrain costs.  We have to wonder why employers, which are assumed to seek continually to minimize input costs and maximize margins, and so to shop for lower prices for all inputs, would not already do so for this particular major cost – which, on their books, is a business expense like any other.  In fact, the literature about the industry, and simple observation, show that employers continually shop and bargain, changing their benefit packages and complaining about costs, while insurers craft plans to meet employers’ concerns about costs (not very well, however).

We also have to wonder how, even if employers cared more about costs, this would enable them to impose better prices and utilization management on the medical world.  Would that give employers and insurers more bargaining power against dominant local hospitals and specialty groups?  Why?  Would it lead to new innovations in efficient delivery systems?  How?  Would it make it easier to generate evidence that could be used to regulate care?  Anyone who thinks so must have missed the recent controversy about mammograms.  The cause and effect connection between the tax treatment of insurance and the delivery of and payment for medical care is long and tenuous.  It requires heroic assumptions to conclude either that employers have not been caring “enough,” given the tax exclusion, or that even if they cared more, they could do much about it just because the tax code changed.

The argument for savings from the excise tax reduces to two propositions.  One says that higher cost sharing will provide “harmless” cost control.  This requires dubious extrapolations from the HIE.  The other is that because of the current tax break for insurance, employers and employees have not cared enough about controlling costs and if they were forced to care more they would be more successful.  This is a remarkable leap of faith about power relations in medical care markets and their consequences.

How to Fairly Address “Excessive Benefits.”

It is reasonable for any society to define a level of medical care that it wants to guarantee to all citizens.  We greatly wish our country would do that.  It is also reasonable and fair to say that coverage beyond that level should not be subsidized, through the tax code or otherwise, except for the poor.  In one way or another, that is what all other rich democracies do.

Defining the standard in terms of the cost, however, just begs for inequity.  It is highly likely to sanction people whose benefits are not unusually generous at all, just more expensive.  The solution is simple: the federal government should define a benefit standard (and, indeed, it will for other purposes under the Senate bill).  Employers and employees should be able to pay for that level of coverage with pre-tax dollars.  Further benefits can be taxable income.  Then the government will not penalize people for being in a risky group, or because health care costs rise more quickly than the CBO’s estimates.  As drafted, the “excise” tax is simply a fall-back cost-control method that targets beneficiaries: if health care costs rise too quickly, the federal government will slash health insurance benefits for people in the employer-based system, even if costs are high because of need for care.

Email This Post Email This Post Print This Post Print This Post

Don't miss the insightful policy recommendations and thought-provoking research findings published in Health Affairs.

Leave a Reply

Comment moderation is in use. Please do not submit your comment twice -- it will appear shortly.

Authors: Click here to submit a post.