Skip to main content

The Cost and Coverage Impact of the President’s Health Insurance Budget Proposals

In this analysis I report the results of my efforts to assess the impact that the President’s health insurance budget proposals on government costs and insurance coverage in the U.S. The results reported here are from the microsimulation model that I developed with the generous assistance of the Kaiser Family Foundation. I also gratefully acknowledge financial support from the Center on Budget and Policy Priorities for this exercise, and the excellent research assistance of Bert Lue.

To summarize, I estimate that the President’s budget proposals will cost almost $12 billion dollars per year if fully phased in. I estimate that these proposals will on net raise the number of uninsured (by 600,000 persons), as those left uninsured through firm dropping of insurance exceed those who gain insurance through taking up tax-subsidized high-deductible plans attached to HSAs.

The Budget Proposal

The President’s budget proposal is focused on promoting the purchase of high-deductible plans attached to Health Savings Accounts, which I will summarize by “HSAs”, primarily in the non-group market. HSAs have two essential features: a high deductible health insurance policy; and a savings account into which individuals can deposit funds tax-free that can be used to fund the deductible, or that can be rolled over for future use. In particular, the President has proposed two mechanisms to subsidize HSA premiums, and one mechanism to further subsidize HSA contributions.

HSA premiums, for purchase of HSAs in the non-group market only, would be subsidized by one of two mechanisms (individuals can choose only one of these two options):

  • Deductibility of HSA premiums, along with a 15.3% tax credit in the amount of those premiums (subject to limits) OR
  • A tax credit of $500/child and $1000/adult, up to $3000/family, that is highly income targeted, phasing out between $15,000 and $30,000 for singles and $25,000 and $60,000 for families.

In addition, HSA contributions, which are already tax deductible, would also be matched with a 15.3% tax credit (subject to limits).

The Model

To estimate the impact of the President’s budget proposals, I employ a micro-simulation model that has been used since 1999 to understand the impact of tax policies (and other insurance market interventions) on public sector costs and insurance coverage in the U.S.[1] This model is very similar to the types of models used by Congressional Budget Office, the Congressional Joint Committee on Taxation, and the Treasury Department in assessing the impact of fiscal policies.

This model takes as its base data on individuals from the Current Population Survey (CPS). Individuals are aggregated into “synthetic firms” by matching with other individuals in the data who match the likely characteristics of their co-workers. Data on health insurance premiums for both employers and individuals in the non-group market are then matched onto these CPS data.

The model takes as its main inputs the details of the policy under consideration, as described above, and then proceeds in two steps. First, it converts the change in policy into a set of changes in prices facing individuals and firms in insurance markets; for example, how much will the subsidies proposed by the President lower the prices of high deductible insurance plans attached to HSAs? Second, it models how individuals and firms will react to those changing prices. For example, how will those who are currently uninsured react to lower HSA prices in terms of purchasing coverage? And to what extent will those with employer coverage leave the employer system to purchase this newly subsidized non-group coverage? The answers to these questions are programmed into the model following the best available evidence in health economics on the responsiveness of individuals and firms to prices changes in insurance markets.

Modeling exercises such as these are fraught with uncertainty, and this is no exception. I have had to make important assumptions as to the attractiveness of HSAs relative to traditional non-group insurance options, and as to the ultimate premiums that will result in a much-broadened high-deductible plan marketplace. I have also not modeled any effect on behavior in either the non-group or employer markets in response to the additional subsidization of HSA contributions or the higher HSA contribution limits. That is, these changes have revenue costs but no behavioral impacts.

The Score

After running the President’s budget proposals through this model, I obtain the results summarized in Table 1. The first column of the Table shows the results for the entire package of proposals, and the second and third columns show the results for each of the premium subsidy options standing alone.


In the first column, I show that the total cost of the President’s proposals, fully phased in, will be $11.6 billion per year, in 2005 dollars. That cost would be lower during the phase in, but would rise over time with health cost inflation. $7.1 billion of that cost is due to subsidizing premiums, and $4.5 billion is due to subsidizing contributions.

The second and third column show the costs of the two “arms” of the premium subsidy. Each has a similar cost, although there are important interactions, as the total cost of premium subsidies of $7.1 billion is well below the sum of these two components.

Take-Up of Subsidies

I estimate that 16.6 million persons will benefit from one or more of these proposals. 8.3 million of those persons will be individuals who previously had non-group coverage (including those already holding high deductible plans attached to HSAs), 4 million will be individuals who previously had employer-provided insurance, and 0.5 million will be individuals who were previously enrolled in Medicaid. Only 3.8 million of the 16.6 million benefiting from this policy (23%) will be those who were formerly uninsured.

As the second and third columns show, the results are quite different for the two different types of subsidies. The universal deduction and payroll-offset credit benefits 14.1 million persons, but only 2.4 million of them (17%) were previously uninsured. The targeted tax credit delivers a much higher share of its benefits to the previously uninsured (29%), but even this figure is quite low.

Impacts on Insurance Coverage

The bottom of the first column of Table 1 shows the overall impact on insurance coverage of these combined proposals. Each part of this bottom panel shows the total change in an insurance category, as well as the flows into and out of that category. For example, I project that the number of non-group insured (which includes HSAs) will raise by 8.3 million persons. 3.8 million of that increase will come from those who were previously uninsured, 4 million will come from those who were previously employer insured, and 0.5 million will come from those who were on Medicaid.

At the same time, I project that the number of individuals with employer insurance will decline by even more, 8.9 million persons. This is because such tax policies significantly erode the existing tax subsidy to employer-provided insurance by equally subsidizing non-employer options. That is, currently employers have a large tax advantage in offering insurance, since wages are taxed but employer-provided insurance premiums are not, and this is part of the reason that so many employers provide insurance to their employees. This policy will remove that tax advantage for HSAs, causing some employers (typically small employers) to stop offering insurance coverage to their employees. Only about half of those employees dropped from employer-provided insurance enroll elsewhere, 4 million in non-group insurance (predominantly tax-subsidized HSAs) and 0.5 million in Medicaid. The remaining 4.4 million become uninsured.

As a result, the number of uninsured persons is projected to actually rise by about 600,000 from these proposals. Once again, this is because the decline in employer-provided insurance of 8.9 million persons is larger than the resulting increase in non-group insurance of 8.3 million persons.

As the next two columns show, this rise in uninsurance is driven by the untargeted deduction/credit, which reduces employer insurance (-7.7 million) by much more than it increases non-group coverage (6.0 million) or Medicaid (0.2 million). The result is a sizeable increase of 1.5 million in the number of uninsured. On the other hand, the targeted tax credit causes a much smaller reduction in employer insurance. This is because this targeted credit benefits a small share of the workforce, which maintains for most workers the tax advantage of employer offering, and thereby leads to much less employer dropping. Take-up of nongroup insurance is smaller as well, but the net result here is a reduction in the number of uninsured of 1.4 million. Once again, interactions between these different pieces of the policy imply that we cannot simply add the estimates from columns (2) and (3) to obtain the overall impact in column (1).


The results of this analysis are similar to the results of my earlier analysis of tax subsidies towards non-group insurance, except that costs are now much higher. As with earlier analyses, I find that the highly targeted credit (point (2) of the approach) would be modestly expensive ($4.2 billion) and would slightly reduce the number of uninsured (by about 1.4 million persons).

But the untargeted deductions and credits raise costs substantially, and on net serve to lower insurance coverage. This is because such untargeted tax policies significantly erode the existing tax subsidy to employer-provided insurance by equally subsidizing non-employer options. This results in small employers dropping insurance, and only about half of those employees who are dropped then enroll in insurance elsewhere (including HSAs).

While this may seem a very large employer response, recall that a reduction in employer-provided insurance is only 5.5% of the total of employer-insured in the U.S., which is actually a fairly modest response to a virtual removal of the tax advantage to employer-provided insurance for those who purchase HSAs.

Table 1:
Simulation Results

  Full Proposal Universal Deduction/Credit Targeted Tax Credit

Cost ($2005 billions)

Total 11.6 4.2 4.0
Premium Subsidies 7.1 4.2 4.0
Contribution Subsidies 4.5 0 0

Takeup (millions of persons)

Total 16.6 14.1 8.3
Previously Non-group 8.3 8.0 4.5
Previously Employer 4.0 3.3 0.9
Previously Medicaid 0.5 0.3 0.5
Previously Uninsured 3.8 2.4 2.4

Change in Population (millions of persons)

Non-Group 8.3 6.0 3.8
From Uninsured 3.8 2.4 2.4
From Employer 4.0 3.3 0.9
From Medicaid 0.5 0.3 0.5
Employer -8.9 -7.7 -2.2
To Non-Group -4.0 -3.3 -0.9
To Medicaid -0.5 -0.5 -0.2
To Uninsured -4.4 -3.9 -1.1
Medicaid 0 0.2 -0.3
To Non-Group -0.5 -0.3 -0.5
From Employer 0.5 -0.5 0.2
Uninsured 0.6 1.5 -1.4
To Non-Group -3.8 -2.4 -2.4
From Employer 4.4 3.9 1.0

End Notes

[1] This model is described in more detail in Jonathan Gruber, “Tax Policy for Health Insurance,” in James Poterba, ed., Tax Policy and the Economy 19. Cambridge, MA: MIT Press, 2005, p. 39-63.