LIFETIME TAXPAYER CONTRIBUTIONS AND BENEFITS OF MEDICARE AND SOCIAL SECURITY

Transcription

1 LIFETIME TAXPAYER CONTRIBUTIONS AND BENEFITS OF MEDICARE AND SOCIAL SECURITY JING GUO and MARILYN MOON Many criticisms of Medicare and Social Security have suggested that beneficiaries get more than they paid into the system over their lifetimes and these entitlement programs may be unsustainable if they are insufficiently funded. But studies do not always use appropriate assumptions for estimating lifetime values. Our simulation model properly aligns the financing sources with the benefits received by making consistent comparisons and using a more appropriate inflation adjustment. Our study suggests that using different assumptions can dramatically change the balance between contributions and benefits as compared with other analyses. Choosing the appropriate assumptions is essential for a reasoned debate on the important issues about the future and financing of Medicare and Social Security. (JEL I13, H5, H24) I. INTRODUCTION Before 1966, the majority of older Americans who needed medical services had to spend their savings, rely on funding from their children, seek welfare or charity, or avoid care. Today, Medicare reduces the risk for large out-of-pocket medical expenditures especially after retirement. Different from most public or private health care insurance programs for working adults, Medicare is an entitlement program that covers most Americans from at least age 65 until death. The Medicare and Social Security programs in the United States provide adequate and affordable protection for all those eligible and additional protections for disadvantaged populations, including frail older adults, disabled, poor, and low-wage workers. Traditionally, the fiscal health and redistributive impacts of those federal programs have been measured on an annual basis. Indeed, the programs were designed to be pay as you go rather than funded programs (in which individuals implicitly have an account to draw upon in the way we think of many current pension systems). However, since taxpayers are the primary contributors to the programs and a taxpayer can Guo: Economist, Center on Aging, American Institutes for Research, Rockville, MD Phone , Fax , jguo.econ@gmail.com Moon: Institute Fellow, Center on Aging, American Institutes for Research, Rockville, MD Phone , Fax , mmoon@air.org be a net receiver in one period and a net payer in another, a lifetime approach is an appealing concept, as people count on these programs to play a key role in retirement and they plan accordingly to invest their money over a lifetime. It has also become popular for taxpayers to question whether they are getting their money s worth in terms of what they pay and what they receive from various programs. Because payroll taxes are paid over a worker s early lifetime and benefits come later, many analysts have evaluated the programs by comparing payroll tax contributions and benefits. Social Security is funded through payroll taxes called Federal Insurance Contributions Act (FICA) tax or Self Employed Contributions Act (SECA) tax. Under Medicare, only Part A is largely funded by payroll taxes, while Parts B and D are funded by premiums paid by Medicare enrollees and general fund revenue. Thus, in this paper, we only focus on the expected value of Medicare benefits for Part A to be consistent with ABBREVIATIONS AIME: Average Indexed Monthly Salary Earning AWI: Average Wage Index CBO: Congressional Budget Office CPI-U: Consumer Price Index for All Urban Consumers FICA: Federal Insurance Contributions Act OASDI: Old Age Survivors and Disability Insurance PIA: Primary Insurance Amount SECA: Self Employed Contributions Act 483 Contemporary Economic Policy (ISSN ) Vol. 36, No. 3, July 2018, Online Early publication March 31, 2017 doi: /coep The Authors. Contemporary Economic Policy published by Wiley Periodicals, Inc. on behalf of the Western Economic Association International. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

2 484 CONTEMPORARY ECONOMIC POLICY the revenues from the payroll tax base when comparing lifetime benefits and contributions. This differs from many previous studies (see, e.g., Steuerle and Rennane 2011; Steuerle and Quakenbush 2012, 2013, 2015). When a worker retires, how much in lifetime tax contributions and benefits for Medicare and Social Security can he/she expect? In recent years, this lifetime contribution measure has been widely reported in the media, and many commentators have pointed to it in discussions of Medicare s future (Moffitt 2016; The National Institute for Health Care Management [NIHCM] Foundation 2015). In addition to the focus on the broad-based average that is commonly reported in policy debates, lifetime measures also capture other dimensions. That is, these programs were designed to be progressive in their impact, that is, to replace a higher percentage of earnings for lower-income workers. Progressivity is usually defined as the degree to which benefits are higher relative to lifetime payroll contributions for lower wage contributors as compared with higher contributors (Biggs, Sarney, and Tamborini 2009; Gustman, Steinmeier, and Tabatabai 2013). Thus, evaluating those programs on a lifetime basis also allows us to consider the progressivity of the two entitlement programs and see how those programs help to reduce inequality. In fact, answering the question of how lifetime benefits vary among different wage earners and cohorts is even more important than the popular focus on assessing an overall average impact. This lifetime approach is not a new measure. The 1998 Bipartisan Commission on the Future of Medicare presented these statistics in its information on financing sources, and it was discussed by various commissioners as influential in their policy proposal development (National Bipartisan Commission on the Future of Medicare 1999). The 1998 estimates for the commission found that, if calculated for those retiring in 1998, expected benefits were substantially higher for nearly all the groups examined, in no small measure because these people would not have contributed anything for about a third of their working lives. 1 The more recent presentations of lifetime contribution measures that have received a lot of attention come from a series by Steuerle and colleagues (2011, 2012, 2013, and 2015). The most attention-getting statistic from their 1. Because collection of tax revenue to fund the program began in 1966 and tax rates were very low for many years after that. findings is that Medicare benefits for the average worker retiring in 2030 will be more than triple their adjusted contributions. This dramatic figure implies a substantial imbalance between contributions and benefits. But these studies overestimate the value of Medicare benefits, biasing the results and making it appear that people do not pay reasonable amounts toward the costs of their care. Moreover, the balance between benefits and contributions varies substantially across income levels. Literature suggests that measurement matters especially when examining inequality (Armour, Burkhauser, and Larrimore 2013). This study demonstrates the importance of using appropriate assumptions for lifetime measurements by evaluating Medicare Part A and Social Security programs from a taxpayer s perspective. We find that high-wage earners contributed more to the tax system than their expected benefits for both Medicare Part A and Social Security, while lowwage workers can expect benefits greater than taxes paid over a lifetime. Thus, we conclude that Medicare and Social Security plays a vital role in reducing inequality because of the progressive nature of the benefits to lower wage workers and are not nearly as imbalanced between taxes and benefits as is commonly claimed. II. METHODS When performing simulations on the lifetime tax contributions and benefits of Medicare and Social Security programs from a taxpayer s perspective, assumptions play a vital role. Steuerle and colleagues (2011, 2012, 2013, and 2015) examined these questions from a similar perspective but used different assumptions. We summarize the key assumptions in Table 1. Areas in which we take the same approach include using typical workers from specific cohorts. A typical worker within each cohort is defined as earning a certain level of wages over a lifetime. In our case, we examine people reaching age 65 in 2000, 2015, and Consistent with the Steuerle approach, we examine the lifetime taxes and benefits by workers at various earning levels: low-wage worker, average-wage worker, high-wage worker, and worker with maximum taxable wage for Social Security ( tax-max wage ). It is critical to consider a range of wage levels to be able to assess the progressivity of the program and to better understand who is being subsidized. An average-wage worker earns the average wage in the economy every year, based

3 GUO & MOON: LIFETIME VALUES OF MEDICARE AND SOCIAL SECURITY 485 TABLE 1 Key Assumptions Domain Previous Studies This Study Direction of Change Common assumptions When to receive benefits i. Medicare: age 65 Single tax payer s earning levels Different assumptions Medicare benefits from a tax payer s perspective Inflation ii. Social security: normal retirement age i. Low wage (45% AWI); ii. Average wage (AWI); iii. High wage (160% AWI); iv. Social Security Maximum taxable wage Part A + Part B + Part D Part A only Medicare benefits CPI-U for both taxes and benefits CPI-U for taxes and Social Security benefits; CPI-Medical Care for Medicare benefits Start working at age 30 till normal retirement age Medicare benefits How long do people contribute Start working at age 22 till Taxes to the system normal retirement age Real rate of return 2% 2%; 3%; 7% Taxes, Benefits on Social Security s estimate from the average wage index (AWI). The low-wage worker earns 45% of the average wage, while the high-wage worker earns 160% of the average wage. The tax-max wage worker earns at the taxable maximum for Social Security every year. Take year 2013 as an example, a low-wage worker s annual salary is $20,308, which is at the 38th percentile rank among all U.S. individuals with incomes; the annual salary of an average-wage worker is $45,129 and is at the 70th percentile rank; 2 $72,206 is the annual salary of a high-wage earner and is ranked at the 86th percentile; and a tax-max wage worker earns $113,700 in 2013 and is at the 94th percentile rank among all U.S. individuals with incomes. Table 2 illustrates the estimated U.S. income percentile rankings of the four wage levels used in this study, using year 2013 as an example. In addition, we assume that a worker always worked at the same earning level without unemployment over his or her lifetime, that is, an average-wage worker always earns the average wage each year. We also assume the same wage structure for men and women, consistent with Steuerle and colleagues (2011, 2012, 2013, and 2015). We do not take wage disparities into account due to the ambiguous line between gender discrimination and preference to work (White House 2015), although women usually earn less than men. Gender differences in our estimates reflect differences only in life expectancy, thus affecting benefits, as described below. 2. This is because the average wage is considerably higher than the median wage, which would reflect the person at the 50th percentile. Consistent with the Steuerle approach, we assume that Medicare benefits reflect an equal insurance value received by each beneficiary in each year. Slightly different, we assume a typical worker will start receiving Medicare Part A benefits at age 65, and Social Security benefits at the normal retirement age of his/her cohort. 3 Thus, we calculate the expected lifetime tax contributions and benefits at the time point of a typical worker s normal retirement age. For the cohort born in 1935, we assume age 65 as the start of Social Security benefits; for the cohort born in 1950, the time point of benefit calculation is age 66; and age 67 for the cohort born in Below, we explicitly contrast our key assumptions with the Steuerle approach (see Table 1) emphasizing four key assumptions: creating a consistent comparison between Medicare benefits and tax contributions, using a more appropriate inflation measure for medical care, adjusting the length of time in the workforce, and using different real rates of return. A. Consistent Comparison between Benefits and Tax Contributions For Social Security, we use the standard assumption that benefits are funded by the payroll taxes dedicated to the Old Age Survivors and Disability Insurance (OASDI) trust funds and that employees bear the burden of both employer and employee contributions. Total lifetime benefits not only depend on how long a person lives, but also the annual expected values 3. Steuerle approach assumes every cohort retires at age 65.

4 486 CONTEMPORARY ECONOMIC POLICY Level TABLE 2 Estimated U.S. Income Percentile Rankings of Wage Levels Personal Income 2013 Estimated U.S. Income Percentile Ranking, 2013 Dollar Amount Among All U.S. Individuals with Incomes Among All U.S. Men with Incomes Among All U.S. Women with Incomes Low (45% AWI) $20, AWI $45, High (160% AWI) $72, Maximum taxable income $113, Data source: U.S. Census Bureau, Current Population Survey, of the benefits. For Social Security, we use the formula for average indexed monthly salary earnings (AIMEs) using the adjusted average wage over 35 years to calculate the primary insurance amount (PIA). For Medicare, the 2.9% payroll tax levied on employers and workers is only used to fund Part A of the program (Hospital Insurance, inpatient care, skilled nursing facility care, home health care, and hospice care). 4 Parts B and D are funded by premiums paid by Medicare enrollees and general fund revenue, rather than the payroll tax. Steuerle and colleagues (2011, 2012, 2013, and 2015) included Parts A, B, and D of Medicare in the lifetime benefits calculation, creating a substantial mismatch between taxes and benefits. In contrast, we restrict the expected value of Medicare benefits to Part A to be consistent with the tax base used, which has a substantial impact on our findings. Today, Part A accounts for only about 44% of total Medicare benefits and covers inpatient hospital care and other institutional services. In our analysis, the annual benefit for a typical beneficiary over age 65 is calculated as per capita average spending under Part A. 5 B. Inflation We use the consumer price index for all urban consumers (CPI-U) to adjust taxes paid to both the Social Security and Medicare programs, and expected benefits received from Social Security. However, when calculating Medicare Part A benefits, we use the Medical Care CPI, rather than 4. Actually some of Part A is funded by taxation of Social Security benefits so even looking at Part A alone can be somewhat deceptive. 5. The expected values of average spending of Part A after 2015 are calculated by projected total spending and total number of enrollments of Part A directly drawn from the 2016 Medicare Trustees Report. CPI-U that others have used. This is because the prices of medical care commodities and medical care services increase faster than other goods in the market and have consistently outpaced the average rate of inflation across the economy (Berndt et al. 2001; Cutler, McClellan, and Newhouse 1998; Newhouse 2001). Using the CPI-U would make benefits appear to grow substantially over time, which is not the case. And, because Medicare benefits could only be used for purchasing medical care rather than other goods, using CPI-U to adjust Medicare benefits as other studies do implicitly assumes that the change in prices of medical care is equal to the change in prices of goods and services across the economy. Thus, using the medical care CPI for Part A Medicare benefits is important for the evaluation and represents a second key reason why our findings differ from others who do not make this distinction. 6 It also means that comparisons across cohorts are more realistic and avoids the impression that younger cohorts will be better off than older ones simply because prices for medical services are higher. Again, this has a substantial impact on our findings. C. Real Rate of Return We calculate the lifetime value of taxes and benefits on the basis of present value at the time point discussed above. This approach allows us to compare the value of money paid or received in 1 year with that of money paid or received in a different year by specific population cohorts. Specifically, this is equivalent to assuming that the taxes paid every year were put into something like a savings account that grows over time until 6. But it should be noted that using this measure of inflation means that taxes and benefits cannot be compared with estimate whether the numbers of actual dollars balance between the two.

5 GUO & MOON: LIFETIME VALUES OF MEDICARE AND SOCIAL SECURITY 487 the taxpayer s normal retirement age. The rate of interest we use to calculate the present value of the lifetime taxes and benefits is the appropriate rate of inflation (CPI-U for taxes and Social Security benefits, and medical care CPI for Medicare Part A benefits, as discussed above) plus a real rate of return. Steuerle and colleagues (2011, 2012, 2013, and 2015) used 2% as the real rate of return in their estimation, which is close to the Congressional Budget Office (CBO) estimate of 2.2% 2.3% in coming decades (CBO 2015). In addition to the 2% real rate of return, we also provide estimates using two alternative assumptions adopted in other literature: 3% (Goldman and Orszag 2014; Nikolich-Žugich et al. 2016; The Office of Management and Budget [OMB] 2003), and 7% (OMB 2003). We illustrate that using lower real rates of return would increase the present value of taxes and decrease the present value of benefits, thus making the two entitlement programs look less generous. D. Age Ranges Appropriate for the Estimates A typical worker in the Steuerle approach is assumed to work every year from age 22 to 1 year prior to the normal retirement age. We use this assumption in one set of our estimates. However, more and more people choose to achieve higher educational degrees and postpone their labor force activities, especially younger generations. For example, about only 40% of adults between the ages of in 2012 worked on a full-time basis (National Conference on Citizenship 2013). Thus, we also provide a set of estimates based on the assumption that a worker begins work at age 30. This alternative assumption provides a likely lower bound estimate of lifetime tax contributions. 7 E. Other Assumptions To determine the number of years of benefit receipt, we use life tables from Social Security for life expectancy by gender starting at normal retirement age. We convert the expected benefit each year to present value until the life expectancy age, and sum the numbers for each year to determine the expected lifetime value of benefits. Because women live longer than men, a typical female worker will have higher lifetime benefits than her male counterpart simply by receiving benefits longer. 7. This would make the two entitlement programs look more generous. We use the 2016 Trustees Reports of both Social Security and Medicare (Board of Trustees 2016; Boards of Trustees 2016) for historical data and projections of future spending. We assume that benefits scheduled in law will be paid even if the trust funds are exhausted. All dollar amounts are adjusted in constant 2016 dollars, allowing us to compare the package of benefits and taxes across generations, for instance, for someone retiring in 2000 with the benefits and taxes of someone retiring in III. RESULTS Our main findings are presented as Figures 1 3, 8 and we focus on Medicare Part A benefits and taxes, which are most affected by our alternative assumptions. For the cohort reaching age 65 in 2030, the lifetime Medicare Part A benefit is $82,000 for a male worker, and $88,000 for a female worker due to slightly longer life expectancy. Assuming the worker starts in the workforce at age 30, low- and average-wage earners would have paid $53,000 and $78,000 in Medicare payroll taxes over their lifetimes, respectively. Workers earning high wages would have paid over $106,000 into Medicare. The lifetime Medicare payroll tax paid by a maximum-wage worker, $189,000, is roughly double the amount of expected Medicare Part A benefit received by seniors in the same cohort. That is, an average-earning male will receive $1.05 in Medicare Part A benefits for every dollar paid in Medicare taxes, and a highwage earning male will only receive 77 cents for every dollar he paid in Medicare taxes. When assuming an earlier workforce starting age of 22 (Table S4, Supporting Information), even an average-earning worker would have paid more taxes than expected benefits from Medicare Part A over a lifetime. 9 The situation is similar for the cohort reaching age 65 in Under the assumption of workforce participation starting at age 30, a high-wage male will receive 83 cents in Medicare Part A benefits for every dollar paid in Medicare payroll taxes, and a max-wage male can only expect 55 cents in Medicare benefits for every dollar he paid in Medicare taxes. When combining both 8. Detailed results are documented in the Appendix, Supporting Information. 9. And since average wage workers are less likely to have delayed labor force entry, this may be a better indicator of lifetime contributions.

6 488 CONTEMPORARY ECONOMIC POLICY FIGURE 1 Net Lifetime Medicare Part A Benefits (Benefits Minus Payroll Taxes) for Single Male, by Cohort Low (45% AWI) Average (100% AWI) High (160% AWI) Maximum taxable wage $29,000 $4,000 $54,000 $20,000 $63,000 $40,000 $18,000 ($24,000) ($16,000) ($3,000) ($66,000) ($107,000) Cohort 2030 Cohort 2015 Cohort 2000 Medicare and Social Security, the workers under the two higher wage categories would still expect less in total benefits than what they paid in payroll taxes for both programs. The cohort reaching age 65 in 2000 has a slightly different story because Medicare did not start collecting payroll taxes until 1966 when the workers in this cohort were age 31, and because the tax rates were significantly lower before the 1980s. In other words, this cohort might have contributed to Medicare taxes for both fewer years and at lower rates than the later cohorts. Thus, when a single male would receive $81,000 from Medicare Part A benefits, many workers paid less than this amount for Medicare taxes over their lifetime, except for the max-wage earners. Regardless of gender, a maximum-wage worker s lifetime tax contribution to Part A of Medicare is $84,000, which is slightly greater than expected lifetime benefits. For all the cohorts examined in this study, we find consistent patterns of net Medicare benefits (lifetime Part A benefits minus taxes, Figure 1). That is, low-wage workers receive positive net Medicare Part A benefits relative to wage contributions, and the net benefits fall as wages rise. Thus, many Medicare Part A beneficiaries will have contributed substantially toward their benefits over their lifetime, and the Medicare program is much more in balance by this measure than the claims often made about the program. The lifetime values of Social Security are much larger than Medicare primarily because of higher tax rates, and our calculations are very similar to previous studies by Steuerle and colleagues (2011, 2012, 2013, and 2015). We illustrate the total benefits (Medicare Part A plus Social Security) and taxes by cohort and different real return rate in Figure 3, taking an average-wage earner male within each cohort as an example. For the cohort reaching age 65 in 2030, expected total benefits are roughly equal to taxes paid toward Medicare and Social Security even using the low real return rate of 2%. The estimated total taxes would exceed expected benefits if using 3% as the rate of real return. IV. DISCUSSION Our evaluation on lifetime contributions and expected benefits indicates that both Medicare Part A and Social Security are progressive from

7 GUO & MOON: LIFETIME VALUES OF MEDICARE AND SOCIAL SECURITY 489 FIGURE 2 Net Lifetime Medicare Benefits by Assumptions, Single Male in 2030 Cohort Low (45%AWI) Average (100% AWI) High (160% AWI) Maximum Taxable Wage $247,000 $221,000 $193,000 $110,000 $67,000 $42,000 $14,000 $30,000 $5,000 $11,000 $(71,000) $(24,000) $(107,000) $(24,000) $(60,000) $(160,000) CPI; Part A+B+D; 2% CPI; Part A; 2% CPI-M; Part A; 2% CPI-M; Part A; 3% a lifetime perspective, fulfilling its intended purpose of subsidizing the benefits of low- and average-wage workers while requiring higher wage workers to pay more into the system than they will draw out. This is a factor that tends to be overlooked in some of the analyses implying that all workers need to pay to cover their later benefits. Our findings differ from others primarily because of two key assumptions regarding Medicare. The first of these differences is in restricting Medicare benefits to Part A. This assumption permits a consistent comparison of how much benefits can be expected from taxes paid. Since payroll taxes don t fund Part B and D, it is simply unfair to consider whether payroll taxes are sufficient to cover a part of the program that other sources fund. The second stems from our inflation adjuster, that is, using medical CPI to adjust Medicare benefits. Because Medicare benefits could only be used for purchasing medical care rather than other goods, using CPI-U to adjust Medicare benefits as other studies do implicitly assumes that the change in prices of medical care is equal to the change in prices of goods and services across the economy, which is not the case. Thus, using the CPI-U to adjust Medicare benefits over time is misleading, and makes them appear to be rising when they have actually changed very little. In Table 3 and Figure 2, we demonstrate that restricting benefits to Part A and using CPI medical care to adjust Medicare benefits result in substantially lower expected lifetime values for Medicare Part A. Applying CPI-U to total Medicare benefits (Part A, B, and D) would nearly triple the estimated lifetime benefits. That is, instead of $82,000 in benefits for a single male, the amount would be $300,000. Taking out benefits from Part B and D drops the estimated level to $120,000, which is still almost 1.5 times our estimate when using only Part A and applying the CPI medical care assumptions. Thus, the lifetime measure is very sensitive to the assumptions. As noted above, we believe these adjustments represent more realistic assumptions about Medicare. In particular, using only Part A is essential for an apples-toapples comparison. And the inflation assumption is critical when comparing benefits across different cohorts. In addition, using 2% 3% as real rate of return is more appropriate, particularly because the estimates extend over a much longer period than most types of lending, and they are more consistent with recent projections (CBO 2015). As presented in Figures 2 and 3, using lower real return rate would increase the present value of taxes and decrease the present value of benefits, making the two entitlement programs look less generous. Thus, using the lower bound of 2%

8 490 CONTEMPORARY ECONOMIC POLICY FIGURE 3 Total (Social Security and Medicare Part A) Benefits and Taxes by Cohort and Rate of Real Return, Average Wage Single Male $1,200,000 $1,000,000 $800,000 $600,000 $400,000 $200,000 $ % 3% 7% Total lifetime benefits Total lifetime taxes TABLE 3 Net Medicare Lifetime Benefits by Assumptions, Single Male 2030 Cohort Assumptions Net Medicare Benefits (Benefits-Taxes) Rate of Real Return Price Index Medicare Parts Wage Level 2% 3% CPI-Medical CPI-U A A + B + D Benefits Low Average High Max-Taxable X X X $299,000 $247,000 $221,000 $193,000 $110,000 X X X $120,000 $67,000 $42,000 $14,000 $(71,000) X X X $83,000 $30,000 $5,000 $(24,000) $(107,000) X X X $72,000 $11,000 $(24,000) $(60,000) $(160,000) real rate of return would provide a likely lower bound estimate for the level of generosity of the programs. Our findings indicate that for younger generations, total lifetime contributions would mostly outweigh lifetime benefits across a range of scenarios. Figure 3 presents the cohort differences by comparing total benefits (Medicare Part A plus Social Security) and tax contributions for an average-wage earner single male within each cohort. Note that for those retiring in 2000, the Medicare taxation did not start until age 31 (in 1966) and the tax rate was low during the first 20 years. Thus, we see this cohort as a more transitional generation than the others. 10 The younger generations would consistently expect to contribute to the two programs more than they would receive. That is, by revisiting this measure using different but more appropriate assumptions, Medicare and Social Security programs do not look as generous in the future as people have sometimes argued. And since most policy proposals for changing Medicare and Social Security 10. This is another reason why using age 30 (rather than 22) as the labor force starting point would provide more comparable results in this study.

9 GUO & MOON: LIFETIME VALUES OF MEDICARE AND SOCIAL SECURITY 491 would only affect these later cohorts, this finding is particularly relevant. Although we limit our key assumptions to emphasize their effects on measurement, some alternative assumptions might be explored in future studies. First, recent evidence suggests inequality in life expectancy by income level (i.e., Waldron 2007). While older Americans have enjoyed substantial increases in life expectancy over time, improvements in recent years have been almost exclusively among those at the top of the income distribution. If we were to apply an alternative assumption of low life expectancy for those with lower wages, both programs would look less progressive. Second, Medicare expenditures might also vary by income level. However, some previous studies provide conflicting evidence on such inequality. For example, Bhattacharya and Lakdawalla (2006) suggested that Medicare spends far more on the poor than the rich at any given age; while McClellan and Skinner (2006) found that Medicare transfers more dollars to high-income neighborhoods. Endogeneity issues would need to be rigorously addressed before applying this alternative assumption. More importantly, since Medicare is a health insurance program, we take the benefits measured as expected value or coverage for the entire risk pool rather than actual individual spending. In addition, existing data that could permit a lifetime approach based on actual spending are seldom available. Third, although women usually earn less than men, we do not take wage disparities into account in this paper due to the ambiguous line between gender discrimination and preference at work (White House 2015). In order to make an appropriate assumption about the gender pay gap in future research, robust evidence on wage disparities by income level is essential. Furthermore, we assume that a typical worker defined in this paper survives to normal retirement age and draws benefits. Because the lifetime perspective adopted in this paper cannot factor in the aggregate effect of taxpayers who never receive benefits (or receive limited benefits), our approach is different from social perspective. That said, we cannot say whether a cohort as a whole will pay for itself, because in that case we would need to weight appropriately the number of people at each wage level and the amount of taxes relative to benefits associated with people at each level to answer that different question. That calculation is something no one has yet attempted. A lifetime perspective on taxes and benefits is just one approach among many to evaluate the social insurance programs, although these programs were never designed to achieve the goal that each generation pays for itself. Medicare was established as a pay as you go system in which current taxpayers fund the program for current beneficiaries. The type of lifetime contribution versus benefit calculations that are now popular adds a new but not singular dimension for judging the programs. Our study suggests that using different assumptions for the lifetime measurements would dramatically change the balance of benefits to contributions and how those advocating change can make claims about the programs. Choosing the appropriate assumptions to make consistent comparisons is essential for a reasoned debate on the future of Social Security and Medicare. REFERENCES Armour, P., R. V. Burkhauser, and J. Larrimore. Deconstructing Income and Income Inequality Measures: A Crosswalk from Market Income to Comprehensive Income. American Economic Review, 103(3), 2013, Berndt, E. R., D. M. Cutler, R. Frank, Z. Griliches, J. P. Newhouse, and J. E. Triplett. Price Indexes for Medical Care Goods and Services An Overview of Measurement Issues, in Medical Care Output and Productivity, edited by D. Cutler and E. R. Berndt. Chicago, IL: University of Chicago Press, 2001, Bhattacharya, J., and D. Lakdawalla. Does Medicare Benefit the Poor? Journal of Public Economics, 90(1), 2006, Biggs, A. G., M. Sarney, and C. R. Tamborini. A Progressivity Index for Social Security. Social Security Administration Issue Paper , Board of Trustees, Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds. The 2016 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds Washington, DC. Accessed June 22, tr/2016/tr2016.pdf Boards of Trustees, Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds. Washington, DC, Accessed June 22, research-statistics-data-and-systems/statistics-trendsand-reports/reportstrustfunds/downloads/tr2016.pdf Cutler, D. M., M. McClellan, and J. P. Newhouse. What Has Increased Medical-Care Spending Bought? American Economic Review, 88(2), 1998, Goldman, D. P., and P. R. Orszag. The Growing Gap in Life Expectancy: Using the Future Elderly Model to Estimate Implications for Social Security and Medicare. American Economic Review, 104(5), 2014, Gustman, A. L., T. L. Steinmeier, and N. Tabatabai. Redistribution under the Social Security Benefit Formula at the Individual and Household Levels, 1992 and Journal of Pension Economics & Finance, 12(01), 2013, 1 27.

10 492 CONTEMPORARY ECONOMIC POLICY McClellan, M., and J. Skinner. The Incidence of Medicare. Journal of Public Economics, 90(1), 2006, Moffitt, R. Medicare s Next 50 Years: Preserving the Program for Future Retirees. The Heritage Foundation. July, Accessed July 31, National Bipartisan Commission on the Future of Medicare. Building a Better Medicare for Today and Tomorrow. Washington, DC: National Bipartisan Commission on the Future of Medicare, National Conference on Citizenship. Millennials Civic Health Index. Washington, DC: National Conference on Citizenship, Newhouse, J. P. Medical Care and Economy Wide Price Indexes. NBER Reporter, 2001, Nikolich-Žugich, J., D. P. Goldman, P. R. Cohen, D. Cortese, L. Fontana, B. K. Kennedy, M. J. Mohler, S. J. Olshansky, T. Perls, D. Perry, and A. Richardson. Preparing for an Aging World: Engaging Biogerontologists, Geriatricians, and the Society. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 71(4), 2016, Steuerle, C. E., and C. Quakenbush. Social Security and Medicare Taxes and Benefits over a Life Time: 2012 Update. Washington, DC: Urban Institute, Social Security and Medicare Taxes and Benefits Over a Life Time: 2013 Update. Washington, DC: Urban Institute, Social Security and Medicare Taxes and Benefits Over a Life Time: 2015 Update. Washington, DC: Urban Institute, Steuerle, C. E., and S. Rennane. Social Security and Medicare Taxes and Benefits over a Life Time. Washington, DC: Urban Institute, The Congressional Budget Office (CBO). The 2015 Long- Term Budget Outlook. Washington, DC: CBO (US Congress), The National Institute for Health Care Management (NIHCM) Foundation. The Budget Deficit and Health Entitlements. Washington, DC: NIHCM Foundation, The Office of Management and Budget (OMB). Circular A- 4, Regulatory Analysis. Washington, DC: White House, Waldron, H. Trends in Mortality Differentials and Life Expectancy for Male Social Security-Covered Workers, by Socioeconomic Status. Social Security Bulletin, 67(3), 2007, White House. Gender Pay Gap: Recent Trends and Explanations. Washington, DC: Council of Economic Advisers Issue Brief, SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section at the end of the article. Appendix S1. Detailed results of estimations