Small Businesses and the Excise Tax
Kaiser Health News and National Public Radio have a jointly-produced story today on small businesses and the excise (Cadillac) tax. It’s reported by Jenny Gold who writes
Higher-cost plans, dubbed “Cadillac” policies by some, often have generous benefits with low deductibles and co-payments, but this is not always the case. Premiums may also be high because customers are charged more because of their age, gender, geographic area or heath status.
Small businesses are at a particular disadvantage. A study published in 2006 in the journal Health Affairs found that the smallest employers pay an average of 18 percent more than large businesses for the same health plan, because they don’t benefit from pooled risk the way a large business does, and administrative costs tend to be higher.
I think there’s more to the story than what was included in the piece. First of all, small businesses will have access to the exchanges. So they won’t be subject to the high loading fees they experience today. Second of all, the risk pool of exchange-based plans will be the entire population in those plans, not the specific population of an employer. Finally, it is my understanding that the excise tax will be modified to accommodate variations in age of risk groups and geographic variation in health care costs.
Putting those reform features together, it doesn’t seem that a small employer has much to be concerned about. In fact, on the whole, I would expect small employers to be better off under health reform. I think the excise tax is only something to fear if one views it in isolation, without consideration of the other reforms that benefit small businesses. But nobody is proposing to pass just the excise tax. So, analysis of it apart from the full package makes little sense and leads to incorrect conclusions.
Is the Cadillac Tax Pigovian?
It seems plausible that most people think health is good, both at the individual and population level. Healthy people are happier, more productive, and more self-sufficient. Good population health obviates public expenditure battling outbreak of disease. Thus, good health is good economics. That is, health has positive externalities. It has social value beyond that which is paid for by and conferred upon the individual.
Economist Arthur Pigou (1877–1959) is credited as the first in the profession to recognize the significance of the difference between private and social value of a good or service. When externalities are (social value is) positive, decreasing the private cost can increase overall welfare. Therefore, decreasing individuals’ cost of health would seem to be welfare enhancing. In contrast, when externalities are negative (think pollution from manufacturing processes), welfare might increase if private costs rise (think carbon tax).
Enter Pigovian taxes or subsidies, named in honor Pigou. A Pigovian tax (subsidy) changes the private cost so that individuals pay (are compensated) for the social cost (benefit). The notion of using the tax system to transfer costs (benefits) of externalites to individuals (also called “internalizing externalities”) is supported by many economists, policymakers, and pundits from across the political spectrum. Many are members of the Pigou Club. Naturally, Pigovian taxes are not universally embraced, and there is also a NoPigou Club.
Given a constant level of taxation, a Pigovian tax system makes sense to me. After all, every tax changes behavior. Holding the overall tax level constant, it is hard to see why it might be undesirable (aside from self-interest) to use taxes to internalize more of the social cost of private behavior. So, count me as a member of the Pigou Club.
Let’s return to health care and health insurance. Under current law, employer contributions to health insurance avoid income and payroll taxation (employee contributions to certain types of plans are also tax exempt). Relative to wages, which are taxed, this is considered a tax subsidy. That’s akin to a Pigovian subsidy for insurance relative to wage and is what one would naively think reasonable given the positive externalities of health. Though if it is reasonable on those grounds why isn’t the tax benefit extended to all types health insurance?
But, health insurance is not health. Nor are health care services. Each may contribute to health but far from efficiently. There are a lot of gears and pulleys between health insurance and health. They don’t run frictionlessly. Not all the ones we pay for connect to health at all. And some of the right levers on health (like good nutrition and exercise) are outside the reach of most health insurance plans. Health insurance is generally poorly designed to promote good health without a lot of waste. (That’s not an argument against insurance. That’s an argument for changing the incentives within the health care system, of which insurance is a part.)
This leads to the question, if the employer based insurance tax subsidy is not a Pigovian one with respect to health, is the Cadillac tax a Pigovian one with respect to insurance? It can only be so if its effect is to cause individuals to internalize some negative externality. What negative externality does tax preferred employer based insurance impose on society?
I confess I can’t think of a slam dunk negative externality. However, it is the case that the relatively rich benefits of tax subsidized employer based insurance promote a climate of excess, unnecessary care and an expectation of access to the latest technology, which in some cases may cost more than the value it provides. In doing so, employer based insurance likely influences medical practice with some spillover effects on Medicare and Medicaid, which are largely taxpayer funded. The cost of overuse of unnecessary care in those programs might be viewed, at least in part, as a negative externality of comprehensive employer based coverage. From this point of view, the Cadillac tax can be seen as a Pigovian one, designed to push some of the social cost of relatively rich coverage back on the individuals who purchase it.
Even if one disagrees with my example of an externality and that the Cadillac tax is Pigovian, there’s reason to support it on other grounds. Reduction of the incentive for provision of economically inefficient services would be one. Equity, considering the size of the uninsured population, would be another. (Aside: I’m not going to rehash all the ways in which the Cadillac tax is poorly designed and could be better. See my prior posts on it for that.)
What is clear is that a Pigovian subsidy on health would make a degree of sense. But it can’t be easily achieved via a subsidy on health insurance for working individuals. There’s just too much play in the connection between insurance and health. If Pigou had a vision for health his gaze probably didn’t settle on taxes or subsidies for employer based insurance. It did, however, focus favorably on the individual mandate.
Cadillac Tax, Redux
Let’s revisit the Cadillac tax using the results presented in my two prior posts on the employer based health insurance tax subsidy. In my first post I showed that the tax subsidy for an individual with a federal marginal income tax rate of 20% and a state marginal income tax rate of 5% is 37%. That is, for each dollar paid by the employer for health insurance instead of wages, government (federal and state) lose 37 cents. That’s the tax subsidy.
Then, in my second post I explained, with the help of Gruber and Lettau, that health spending increases by 0.7% for each 1% drop in cost of insurance (that’s the -0.7 elasticity on the intensive margin I described). Putting that together with the 37% tax subsidy, we find that health spending by Americans with employer-based insurance is higher by 26% (=37% x 0.7) than it would be without the tax subsidy. Note that this assumes an average federal marginal tax rate of 20% which, if anything, is a bit low.
Let that sink in for a minute. Health care expenditure by workers and their families with employer based insurance is 26% higher due to the tax subsidy (a follow-up post attempts to convert this to an annual dollar figure). That’s huge! If we could beat that back, we’d make a substantial dent in the degree to which we overspend (relative to other OECD countries) on health care. And that’s precisely what the Cadillac attempts to do, in its special and blunt way.
Now, I’m not a fan of the precise structure of the Cadillac tax. It could be improved in many ways, e.g. adjusted for geographic variation in health care costs and risk profile of workers. Another thing I’d like to see is that it be made more progressive by setting it to be the actual worker-specific tax subsidy rate rather than a flat 40%. But to do that would require imposing the tax directly on workers instead of on insurance companies. Since workers will ultimately pay the tax either way, the politically expedient ruse that insurers are the ones being taxed actually prevents the Cadillac tax from being more sensibly designed. That’s a shame.
So, if you don’t like the Cadillac tax because it is crude and imprecise, I’m with you. But the basic idea of ratcheting back the tax subsidy is sound on economics grounds. I just wish we could do it better, though I also understand there are political obstacles to a sound design. The Cadillac tax is an attempt at a solution to both a policy problem and a political one. Unfortunately, it solves neither perfectly. Few products of the legislative process do.
Understanding the Employer Based Insurance Tax Subsidy
The economist’s knock on employer based insurance is that it is paid with pre-tax dollars. That means that each dollar of worker compensation in the form of health insurance is worth more than a dollar in compensation in the form of wages. One consequence is over consumption of health insurance and health care. The tax subsidy also contributes to job lock and other labor market distortions because it makes employer based insurance more valuable than non-group health insurance.
So, the decks are tilted toward employer based insurance but by how much? What is the tax subsidy per dollar of insurance? It is tempting to say that it is the worker’s federal marginal income tax rate. But that’s not right. It is not even close. The key is to the right answer is to ask, for each additional dollar an employer pays toward its wage bill, how much does the employee receive in after-tax compensation? That is, your employer paid a dollar toward covering your wage. What portion of that dollar showed up on your net paycheck?
Let’s start with the additional cost to the employer due to the Social Security payroll tax rate (tSS) and the Medicare payroll tax rate (tMC), ignoring the high-income cutoff of the Social Security tax. Each dollar of gross wage costs your employer (1+tSS+tMC) dollars. Or, put another way, for each dollar your employer pays toward its wage bill on your behalf you only receive 1/(1+tSS+tMC) dollars. But that’s before you pay your taxes. So, let’s do those next.
The final dollar of gross wage is taxed at your federal marginal income tax rate (tF), your state marginal income tax rate (tS), and is also hit with Social Security (tSS) and Medicare payroll taxes (tMC). Therefore, what you actually receive post-tax from that dollar of wage is 1-tF-tS-tSS-tMC of a dollar. Hence, for the marginal dollar of employer cost, you actually receive post-tax
TP = (1-tF-tS-tSS-tMC)/(1+tSS+tMC).
The quantity TP in the foregoing expression is known as the tax price and has played an important role in theoretical and applied economic analysis of employer based health insurance. (See the fifth page (marked 297) of The Impact of the Tax System on Health Insurance Coverage by Jon Gruber.) The amount by which TP differs from one is the cost avoided when the marginal dollar of compensation is provided in the form of health insurance instead of wage. Thus 1-TP is the amount in taxes lost to government. Already we can begin to sense that this is quite a different animal than the federal marginal tax rate.
Let’s plug in some numbers. Suppose your federal marginal income tax rate is 20% (selected to correspond to James Kwak’s calculation), your state marginal income tax rate is 5%. The employer and employee Social Security payroll tax rate is 6.2%. And the employer and employee Medicare payroll tax rate is 1.45%. Plugging these in (as decimals, not percentages) to the tax price equation above we find that
TP = (1-0.2-0.05-0.062-0.0145)/(1+0.062+0.0145) = 0.63.
Thus 1-TP is 0.37. Even though your federal marginal income tax rate is only 20% government (federal and state combined) loses 37 cents of tax revenue for each dollar paid in health insurance as opposed to wage.
This is why the tax subsidization of employer based health insurance is a big deal. It goes way beyond the federal marginal tax rate. Assuming that’s all there is to is a mistake (h/t Tyler Cowen; note also Henry Aaron’s correction). Finally, a Cadillac tax of 40% is pretty close to what would be required to recover the lost tax revenue for the example above. But it wouldn’t be enough for individuals with a federal marginal income tax rate above 20%. In this sense, the Cadillac tax is a bargain.
Later: See also my follow-up post on another important fact about the employer based tax subsidy.
Gruber, the Excise Tax, and the Politics of Policy
A lot of well-meaning people and good ideas can get tarnished when they enter the realm of politics. Take Jon Gruber, who’s been in Krugman’s words, the “go-to guy on modeling reform” and a “top micro-modeling expert”. Due in small part to the fact that he didn’t disclose his contract with HHS to all people all the time, and in large part to bloggers at firedoglake he’s been accused of harboring conflicts of interest.
Sometimes a kerfuffle is just a kerfuffle and not worthy of a lot of worry. Again, Krugman:
The truth is that this is no big deal. Gruber’s grant is from HHS, not the West Wing; it’s basically the same kind of thing as, say, an epidemiologist receiving a grant from the National Institutes of Health. You wouldn’t ordinarily say that this tarnishes the epidemiologist’s credentials as an independent analyst on infectious diseases…
The only reasons you might see this differently would be if Gruber were either receiving a sweetheart deal, or seemed to have changed his views to accommodate his sponsors. Neither is remotely true.
Speaking of Gruber and small kerfuffles, he’s one of the several individuals called to task by Larry Mishel in his piece on the historical trends in health care costs and wages. But in Gruber’s case, I think this is making a mountain out of a molehill. Though Mishel has a narrow point I agree with, that wage changes aren’t entirely explained by those of health care costs, I don’t think Gruber’s words in his 28 December 2009 Washington Post Op-Ed explicitly contradict that point (though one could read his implication that way).
Moreover, the debate over the extent to which changes in wages can be related to premiums is silly, because we know the answer. In the long-run it is one-to-one. Workers pay the premiums even if employers appear to and even if those payments aren’t the main or only cause of wage changes over the last two decades. It misses the point of the excise (Cadillac plan) tax to in any way obscure or avoid that fact. So, in my view, it is the fact worth emphasizing: workers pay the premiums.
Workers will also pay the excise tax on premiums. That fact is receiving not one, but two levels of obfuscation in the policy debate. John Kerry is characterizing the excise tax as one that won’t harm workers because it is levied on insurers (h/t Ezra Klein). But insurers will bill their costs to employers who pay the premiums (at least in part). Employers will pass that cost along to employees, through–that’s right–lower wages. Workers pay the premiums.
Of course the whole point of the excise tax is to raise revenue in a way that also promotes efficiency in the health care system. It will certainly do the former but it is a very imprecise and flawed way to do the latter. It’s regressive relative to removing the premium tax exemption. In the form publicly proposed to date it makes no accommodation for geographic variation in health care costs or in the risk profile upon which the premium is based. Put simply, in implementation it’s a stinker (though that could be fixed) with its main redeeming quality its intent (which is good).
Nevertheless, it is being defended as a way to begin to remove the special status that employer-based health care premiums receive by the tax system. Again, this is a defensible and laudable goal but the excise tax would replace one flawed system for another. And I have to believe that the smart people defending it know all this. So why is it being defended so strongly? The only plausible answer I can come up with is that it is, somehow, good politics. Or, at least, it is less bad than alternative ways to raise revenue for health reform.
When good people (like Jon Gruber) and good ideas (like eliminating the special tax treatment of health care premiums) meet politics their intent and image can get distorted. It’s a sad fact that the surest way to stay pure is to stay away and to keep your good ideas quiet. But that doesn’t help anybody but yourself.
Thoughts on the Cadillac Tax
Aaron Carroll sent out the beacon. He quotes the abstract of a recent Health Affairs paper:
It’s often assumed that high-cost health insurance plans—sometimes called “Cadillac” plans—provide rich benefits to plan subscribers. Health reform provisions that treat these plans like luxuries may be misguided. Only 3.7 percent of variation in the cost of family coverage can be explained by benefit design (actuarial value). Benefit design plus plan type (HMO, PPO, POS, or high-deductible plans) explains 6.1 percent of this variation. Industry type and medical costs in the region also play a role. Most variation in premiums, however, remains largely unexplained.
Then he muses
I’m not sure how people still envision the excise tax working to reduce costs.
I need a health care economist to provide a reasoned take on this peer-reviewed study and explain why it’s wrong, or how the excise tax will function in light of its findings.
I’m looking at you, Austin.
One of the embarrassing limitations of the glorious institutions with which I am affiliated is I cannot access newly released online Health Affairs articles. So, if I’m to review the one Aaron references he’ll have to send me the PDF.
However, let’s presume the results are reasonable. Then there would seem to be an awful lot of variation that has nothing to do with excessively lavish benefits (due to regional pricing differences, practice patterns, policyholder risk, etc.).
The first question I would ask is: rather than look at variation in premium levels, to what extent does generosity and plan type explain an indication of being above or below the Cadillac tax threshold? Perhaps there is a lot of variation in level not explained by those things but those things do a good (or better) job of predicting whether or not the plan is an outlier. Put another way, the authors don’t seem to have hit the nail on the head here, though they are very very close (and having not read the paper maybe they have another model in there that didn’t make the abstract).
Second, I don’t think the Cadillac tax is well designed in many respects. Some of these have been discussed by others (Klein and Volsky). My personal take on its limitations is that it is not a measure of premium affordability, which is what it should be, because it leaves out income. See my post on Regulated Compensation.
Having said all that, in theory anything that is taxed should experience a lower demand. Therefore, one would expect that individuals and employers will seek lower premium plans to avoid the tax. In doing so they have to give up something and it can only be benefits or show up in the form of increased cost sharing. That could induce lower utilization and therefore lower costs overall. Perhaps insurers will try harder to create plans with lower premiums (to woo employers). One way they could do so is by negotiating lower rates with providers if they can. Economic theory would suggest they already get the lowest possible rates given market power. So I don’t know that there is much to gain there.
On the whole, the tax is quite a ways removed from the real problems, which are over-utilization of low efficacy care and high unit prices. Were I to design a tax I’d want to target those things.
The Cadillac tax is blunt but it raises revenue while avoiding the politically difficult fight with providers. That seems to be what the White House and Democrats are seeking (at least Senators). I’m willing to accept putting off the real tussle over health care costs for now just so we can achieve some important insurance market reforms and set up some structures (do some pilots and demos) to prepare for that fight later, which is what current legislation would do. But someday, and quite soon–but not this year–we’ll have to return to the cost issue in earnest.
The Cadillac tax is a stop gap measure, but nobody should fool themselves into thinking it will solve the real cost problem. Meanwhile it seems to be irritating a lot of people so I’m beginning to wonder if it is as politically savvy as it may have once seemed.
Subscribe
