Distribution of Lifetime Medicare Payroll Taxes and Spending By Lifetime Household Earnings

Growth in federal spending for Medicare—the largest federal health care program—is expected to outpace the economy in the coming decades, prompting policymakers to consider measures to slow the program’s growth. The consequences of such measures for beneficiaries in different socio-economic groups depend on the distribution of Medicare taxes and spending under the current system. However, only a few academic studies have estimated that distribution, and they offer conflicting views. This distributional study incorporates new estimates of the distribution of Medicare spending across beneficiaries with different levels of lifetime household earnings.

A priori, the distributional effects of the Medicare program—which depend on how contributions paid into the system and costs to the system to provide benefits vary among population groups—are ambiguous. Contributions come mainly from premiums, dedicated taxes, and general revenues; costs mostly consist of payments for services covered by Medicare, and they also include other payments made directly to the providers and administrative expenses. While premiums are similar across beneficiaries, we expect higher payments of dedicated taxes from beneficiaries with higher lifetime household earnings. (This study does not consider general revenues.) We also expect lifetime spending for benefits to vary across beneficiaries because of variation in both annual spending and life expectancy. On average, beneficiaries with higher lifetime household earnings might have lower lifetime spending because of better health status resulting in lower annual spending. Alternatively, they might have higher lifetime spending because of higher annual spending associated with greater demand for health services or longer life expectancy.

Our analysis of the distributional effects of Medicare uses a unique dataset with information on beneficiaries’ lifetime earnings and Medicare spending in addition to demographic projections from CBO’s long-term microsimulation model (CBOLT). We first estimate the relationship between lifetime household earnings and annual Medicare spending using the 1994-2005 National Health Interview Survey linked to Medicare and Social Security Administration (SSA) administrative files. We use the Primary Insurance Amount (PIA) from the SSA files—a measure of lifetime individual earnings—to proxy for lifetime household earnings. We then incorporate the estimated relationship between PIA and annual spending into CBOLT to project lifetime Medicare spending for each individual in the model: We allocate total Medicare spending in each year to beneficiaries on the basis of the average cost per beneficiary, which depends on the beneficiary’s PIA and other characteristics. To illustrate the distributional effects of Medicare, we present lifetime Medicare spending (net of premiums) and lifetime dedicated taxes by quintile of lifetime household earnings for beneficiaries born in the 1950s and later.