Sunday, December 05, 2004

The Tax Treatment of Health Insurance

Brad DeLong is disappointed in Jonathan Weisman's recent Washington Post story on likely directions for economic policy during Bush's second term, motivated at least in part by CEA Chairman Greg Mankiw's speech at the American Enterprise Institute last week. I'm going to guess that the offending paragraph, in Brad's view, is this one:

The argument points to a certain truth about President Bush's free-market economic policies that Bush supporters say is unappreciated: In crafting a broad agenda for his second term, Bush is trying to adhere strictly to economic theory, perhaps even more so than during the Reagan administration's early battles over deregulation and taxes.
Brad goes on to point out elements of economic theory that would argue against removing the tax deductibility of employer-sponsored health premiums. I'll get to those in a minute.

I don't recall having met Jonathan Weisman while in DC, but I spoke to enough reporters to know the sort of assignment he may have had in writing about this: Mankiw gave a high-profile speech, in which he laid out some key issues and linked them with theory; so go get some alternative viewpoints and try to tie them together in a story to keep the discussion going. There are bound to be problems with a story that tries to string together viewpoints rather than, say, evidence. I don't envy the task of writing such a story--it's neither art nor science.

What Brad (and I) would like reporters to be good at is doing something like this, the first entry of which would lead to this paper by John Sheils and Randall Haught, which estimates the total cost of providing the tax-deduction to employer-sponsored health premiums and its distribution by income. In addition, we'd like for the reporters to know where to look for solid economic research, for example, by keeping up with this publication every month. Searching that page for the words "health insurance" leads to a summary of this paper by Jon Gruber and Ebonya Washington on the extent to which tax subsidies induce people to purchase health insurance.

Now we're ready to start talking about how to craft policy with an eye toward economic research. The question at hand is whether we can positively affect the market for health care by removing the tax-deductibility of premiums, or, more specifically, the differential tax treatment of premium relative to out-of-pocket expenses. The fact that premiums are deductible while out-of-pocket expenses are not (if they are not reimbursed by a flexible spending account) causes health insurance plans to have high premiums and generous first-dollar coverage of health expenses. At the margin, individuals and their doctors do not pay for the treatments that are undertaken. This should cause more treatments to be undertaken. This is the economic theory part. It is recognized by all of the sources in the article, and it is the "moral hazard" problem listed as #3 on Brad's list.

There are two ways to remove this distortion--by allowing both types of expenses to be deducted or by allowing neither type to be deducted. The introduction of Health Savings Accounts (HSAs) in the Medicare bill last year was a way to do it based on the former. I favor the latter. Here's why.

The deduction exists because there is a notion that the government ought to encourage people to get health treatments that they need. Deductibility is a terrible way to do this, on equity grounds. Take a look at Exhibit 1 in the paper by Sheils and Haught. It estimates that $209.9 billion of tax receipts are foregone in 2004 because of deductibility, with $188.5 at the federal level and $21.4 billion at the state level. Of the federal portion, $101.0 billion is due to the income tax deductibility, with another $66.4 billion due to exemption from the payroll tax for Social Security and Medicare. That's one large chunk of change.

Who benefits from this deductibility? Exhibit 2 in their paper shows that the average family with $100,000 or more in income receives a benefit of $2,780. Compare this to an average benefit of $1,231 for a family with $30,000 - $39,999 in income. Because tax rates are higher at higher income levels, and those with higher incomes are more likely to have coverage, the benefit goes up with income. Exhibit 3 in their paper estimates that, in the aggregate, the 14 percent of the families with incomes over $100,000 receive 26.7 percent of the federal tax benefits, compared to 28.4 percent of the benefits received by the 57.5 percent of the families with incomes below $50,000.

The portion of this disparity that is due to the progressivity of the tax system is ridiculous. Subject it all to tax, and take some portion of the $100 - $200 billion saved and use it to provide refundable tax credits to purchase health insurance, whether through an employer or an individual policy. The credits should phase out at higher income levels. (Credits offset a tax liability dollar-for-dollar. Deductions offset taxable income dollar-for-dollar, and are those more valuable to people who pay higher tax rates on that income.) This is a far more equitable way to use tax revenues (or their absence) to promote health insurance coverage, in addition to its efficiency consequences for reducing moral hazard.

Brad's concern, which is expressed in the other 4 pieces of economic theory that he says are relevant for the article in the Post, pertain to possible consequences of making such a shift. He writes:

The other four principles of economic theory strongly suggest that trying to push the country out of its current pattern of health-care financing into one in which individuals bargain one by one with insurers for their coverage would be a very bad idea.
It is worth pointing out that replacing deductions with credits as I have described does not necessarily push us to a situation where there is no group coverage. In fact, it may not be appreciably less of a group market than it already is. Would Dartmouth (or Berkeley), for example, now drop its group health coverage? Not a chance. Would the lack of a marginal subsidy to purchase health insurance cause the ranks of the uninsured to swell? Not if I am reading the paper by Gruber and Washington correctly. The authors estimate that a 10 percent higher tax subsidy for premiums increases voluntary take-up of health insurance by only 0.2 percent. This is a tiny response, and it seems reasonable to assume that it would also apply to a policy change that lowered a tax subsidy as well.

With the tax deductibility eliminated, we would have a more progressive means of offsetting health insurance costs across families and a more efficient means of delivering the care, since employers would have good reason to substitute toward plans in which employees pay more of their costs at the margin, for a given average cost.

Other blogs commenting on this post


Adam O'Neill said...

Gruber and Lettau's paper "How Elastic is the Firm's Demand for Health Insurance?" (NBER WP 8081) provides estimates of the elasticity of insurance offering with respect to the net-of-tax rate that are more relevant for assessing the effects of tax reform on insurance coverage than are those in the Gruber and Washington paper. Their estimate is considerably higher (-.31 to -.41). As they write "Our simulation results suggest that major tax reform could lead to an enormous reduction in employer-provided health insurance spending."

You are right, however, that deductions for employer-provided health benefits are regressive. [More generally, so are all tax deductions, including those for mortgage interest and state income tax.]

Anonymous said...

I don't think it is reasonable to assume that the response to an increase in the subsidy (Gruber/Williams) will be symmetric with a removal of the subsidy. The response is likely to be asymmetric for at least two reasons:

1) the risk averse and well to do have already chosen to accept the health insurance offer, leaving only the poor and risk neutral to respond to the subsidy. Yet the whole population will respond to the removal of the tax advantage (Bush proposal)

2) Kahneman and Tversky have spent a lifetime demonstrating that people respond asymmetrically to different types of losses (see "Prospect Theory"). Additionally, they also demonstrate how people incorrectly assess small risks. Consequently, the removal of the subsidy will be percieved as a large utility loss.

Finally, (and I'm surprised that Brad didn't mention it), the idea that MB=MC in the healthcare exchange is hard to fathom. Who can reasonably assess when the cost of a treatment outweights the benefit? Isn't the inherent information asymmetry the reason we go to the doctor?

Anonymous said...

"The portion of this disparity that is due to the progressivity of the tax system is ridiculous. Subject it all to tax, and take some portion of the $100 - $200 billion saved and use it to provide refundable tax credits to purchase health insurance, whether through an employer or an individual policy."

And back here on planet earth, the Republican majority would actually "subject it all to tax, and take some portion of the $100-200 billion saved" to reduce taxes on capital gains and corporate dividends...

It's good to see that a Republican economist has some concern for the regressivity of federal policy, but I tend to think you're part of a very small minority.

OHenry said...

This post has been removed by a blog administrator.

NHS Furniture said...

Top blogging. Very interesting post

website marketing said...

please continue your blog!