A Framework for Medicare Reform

Studies | Health

No. 315
Friday, September 12, 2008
by John C. Goodman


Estimating the Effects of the Reforms

Although we have considered a large number of reforms in the preceding section, here we attempt to simulate the effects of just a few of those reforms. We begin with the core model, described above. We also consider two other reform proposals.

Simulation Assumptions: The Core Model. In order to simulate the effects of our proposed reform, we make the following assumptions [see Figures II and III]:

  • Future health costs will grow at the same rates assumed by the Medicare Trustees. Note: This is a conservative assumption, since it ignores the effects of increased incentives we propose to control costs.
  • All current Medicare enrollees opt to enroll in SCP plans. This assumption is made for ease of computation and has a very minor impact.
  • All future participants enroll in SCP plans at age 65.
  • The mandatory contribution is 4 percent.
  • All HIRA funds are invested in a balanced portfolio of assets earning 5.2 percent during the accumulation phase and 2.9 percent (for the annuities) or 5.2 percent (for programmed withdrawals) during the decumulation phase. This capital market return is lower than what others have suggested.38
  • The base SCP plan begins in 2007 with a $2,500 deductible and is indexed to Medicare per capita cost growth.
  • The demand-side and supply-side effects of higher deductibles are modeled based on the results of the RAND Health Insurance Experiment and other studies.39
  • The government makes an annual contribution which, when combined with a 15 percent premium payment and an individual’s HIRA annuity or programmed withdrawal, is sufficient to pay the risk-rated premium needed to purchase an SCP plan for every retiree.
  • If the HIRA balances of some individuals become so large that no government contribution is needed to purchase an SCP plan, the government taxes the surplus funds of some individuals in order to make risk-adjusted payments for others.

An Alternative Way to Purchase Private Health Insurance: The Rettenmaier-Saving Plan. Andrew Rettenmaier and Thomas Saving have a proposal that we model below.40 [See Figures IV and V.] Specifically, the retiree’s entire HIRA annuity is deposited in the Roth-type HSA and the retiree’s across-the-board deductible increases dollar for dollar with the HIRA annuity amount. For example, if the HIRA annuity is $1,000, the total deductible would rise from $2,500 to $3,500.

The advantage of this approach is that it generates more demand-side incentive effects for every retiree and, as a consequence, it generates more supply-side effects as well. Therefore, overall health care spending will be lower than otherwise and the retirees will be able to devote any of their annuities that are not spent on health care to the consumption of other goods and services. Another advantage is that workers will not view their contributions as a pure tax — as they would with the core reform that requires the entirety of their annuities be devoted solely to health care consumption. The downside of this approach is that it leaves taxpayers with a higher long-run burden than the core model. However, retirees are able to consume more of other goods and services from the residual amounts of their HIRA annuities that are not spent on health care. This higher consumption is not reflected in the following simulation results.

The assumptions for the Rettenmaier-Saving Model are the same as for the Core Model except that:

1. HIRA annuities (or programmed withdrawals) are deposited directly into the recipient’s HSA and the across-the-board deductible rises dollar for dollar with each contribution. The annuity combined with the indexed-based deductible defines the upfront cost sharing for each retiree.

2. Individuals have a property right in their entire HIRA annuity, regardless of income or health status.

Another Way to Purchase Private Health Insurance: The Intermediate Model. Under the Rettenmaier-Saving proposal, accumulations in HIRA accounts can become quite large over time, especially for high-income workers. In some cases the annual annuity payment could even exceed per capita Medicare spending. The incentive effects of a high deductible, however, are largely captured by the time it reaches $5,000. For this reason we consider an intermediate model in which the maximum deductible is $5,000. [See Figures VI and VII.] This amount is indexed by per capita Medicare spending. Retirees face cost sharing up to the maximum deductible and retain any unspent HIRA annuity deposits. If an individual’s annuity amount is greater than the difference between the $5,000 maximum deductible and the base deductible, the excess is used to pay for general Medicare spending on all beneficiaries.

The assumptions under the Intermediate Model are the same as for the Core Model except that:

  1. HIRA annuities (or programmed withdrawals) are deposited directly into the recipient’s HSA and the across-the-board deductible rises dollar for dollar with each contribution up to a maximum deposit of $2,500 and a maximum deductible of $5,000 (with amounts indexed to per capita Medicare expenditure growth).
  2. Individuals have a property right in their entire HIRA annuity up to the indexed value of the maximum deductible, regardless of income or health status.

Simulation Results. Figures II through VII show the simulation results for the three models and the two rate-of-return assumptions during the decumulation phase. Note that in each of the graphs:

  • The upper line represents Medicare spending (funded by taxes plus premiums) under the current system.
  • The next line down represents Medicare spending under the reformed system (funded by taxes plus premiums plus prefunding through HIRA accounts).
  • The next line represents the contemporaneous cost of the reformed program (measured as the sum of taxes plus premiums needed to pay benefits in each year) plus contributions to HIRA accounts needed to continue the process of prefunding future health care expenses.
  • The bottom line represents only the contemporaneous cost (taxes plus premiums) and is the one most comparable to the top line.

As all six figures show, the effects of prefunding combined with better incentives have a dramatic impact on the future financial health of Medicare. Eventually, the taxes and premiums needed to pay benefits comparable to what Medicare promises today will be only a fraction of what they will need to be in the absence of reform. Specifically:

  • Under the current system, the taxes and premiums needed to support Medicare spending will more than triple, growing from 3.2 percent of GDP today to 11.3 percent by 2080.
  • By contrast, under the Core Model with a 2.9 percent decumulation rate of return, the taxes and premiums need to support Medicare spending will be 2.9 percent of GDP by 2080.
  • Under the higher (5.2 percent) decumulation rate of return, the burden of Medicare will return to its current level soon after midcentury and it will be only one-third of the current level by 2080.

The burden of Medicare spending achieves comparable results under the Intermediate Model.

Although both models bring the burden of Medicare spending back to current levels, by the time today’s teenagers reach the age of retirement the Intermediate Model has a bigger impact on health care spending, bringing the total down to 8.2 percent of GDP as opposed to 10.2 percent with the Core Model. Relative to the Core Model, the Intermediate Model cuts spending by one-fifth. Figures IX and X show how the burden for taxpayers will change over the next 73 years: Whereas taxes pay for almost nine of every 10 dollars of Medicare spending today, under the Intermediate Model taxes will be needed for only one in every four dollars of spending in 2080.

As explained in the Appendix, we assume that the Core Model has demand-side effects only. For the Intermediate Model, however, we estimate supply-side effects as well:

  • By 2080, with a 2.9 percent return during the decumulation phase, the Intermediate Model will reduce the burden of Medicare by 10.2 percentage points of GDP, relative to what otherwise would have happened.
  • Of the total reduction, about 70 percent (7.1 percentage points) is produced by the effects of prefunding, about 20 percent (2.1 percentage points) is produced by demand-side effects, and about 10 percent (1.0 percentage points) is produced by supply-side effects. [See Figure VIII.]

Consider now the Rettenmaier-Saving reform plan (illustrated in Figures IV and V), which allows rising HIRA contributions through time to be contributed to the retirees’ HSA accounts along with dollar-for-dollar increases in the across-the-board deductible. The results for the 5.2 percent/2.9 percent return assumptions are as follows:

  • This reform substantially reduces Medicare spending — falling from 11.3 percent of GDP in 2080 to 5.7 percent — virtually cutting total spending in half.
  • The remaining burden of taxes and premiums, however, is 70 percent larger than under the Intermediate reform (5.7 percent versus 3.3 percent, with a 2.9 percent decumulation rate).
  • The total cost of the reform is 7.4 percent of GDP in 2080 including workers’ contributions to their HIRA. This compares to 5.0 percent of GDP under the Intermediate reform.

The reason for this apparent anomaly is that HSA account-holders are choosing to spend substantially more of their HSA funds on other goods and services (rather than health care) than is allowed under the Intermediate plan.


Read Article as PDF