Who Pays for Social Security —

(and Who Ought To)

syl schieber served as commissioner of the Social Security Advisory Board. A more detailed technical version of this analysis is available in the Fall 2019 issue of the Journal of Retirement.

Illustrations by pep montserrat

Published January 23, 2020


Earlier this year, William Gale of the Brookings Institution wrote about saving Social Security in the Milken Institute Review, laying out a number of objectives such an effort should embrace.

I agree with several of these:

• reform should eliminate the program’s financing shortfall and put it on a sustainable path forward.

• the money should come from a dedicated source of revenue.

• any cost-cutting should protect the most vulnerable.

• reform should reflect the realities of the changing market for labor.

But Gale offers three other objectives that, in my view, require critical assessment. Specifically, he suggests reforms should (a) not reduce the incomes of those near retirement or currently retired; (b) respect public opinion that favors higher taxes over reduced benefits; and (c) address Social Security’s financing shortfall sooner rather than later. Think of this article as a reconsideration of some reform ideas that have gained great currency among center-left policy thinkers.

Putting Generational Equity in Context

Many historical analyses of Social Security note that early generations in the system received benefits of greater value than their lifetime contributions, thus creating legacy obligations to be paid for by subsequent participants. There are varying estimates of when the “too-good-to-be-true” party ended—when new beneficiaries moved from being net lifetime gainers to net lifetime payers. But most suggest that point had been passed by early in the 21st century. The challenge is how to distribute these accumulated legacy costs across current and future participants in the system.

Gale suggests that “going forward, workers should pay enough to cover the benefits of their own generation plus some of the legacy burden.” I believe this perspective is too narrow because it lets current beneficiaries and those who will be coming online over the next 10 to 15 years completely off the hook, freeing them from any responsibility for liabilities that have shadowed the system for much of their working lives. The fairness problem is compounded by the fact that those who would be exempted are largely baby boomers, by far the largest generation ever to claim Social Security benefits.

One of the issues in implementing a policy of requiring workers to pay enough taxes to cover the costs of their own generation’s retirement is defining just what that means. Traditionally, the Social Security pension system has been considered fully financed if projected revenues were sufficient to cover projected costs over the next 75 years. But developments since the early 1980s offer good reasons to believe that perspective is inadequate—that adequacy should also mean that the projected financing be stable at the end of the period. Officially, however, 75 years is still the period for assaying financial adequacy, technically referred to as being in “close actuarial balance.”


Our knowledge that Social Security is inadequately financed dates back more than a quarter-century, and doing nothing about it over the intervening years has resulted in a redistribution of costs across generations. The Social Security Trustees were aware the system was facing shortfalls under the close actuarial balance test by the late 1980s. In annual reports to Congress from 1988 through 1990, they removed the usual confirmation of this status. Indeed, Social Security’s chief actuary formally acknowledged this failure at the end of the 1990 Trustees Report. A year later, even after changing the definition of close actuarial balance, the Trustees stated, “the OASDI program is not in close actuarial balance in the long range.” (OASDI is Old Age, Survivors and Disability Insurance, the catchy bureaucratic name for Social Security.) Similar statements have been included in every subsequent annual Trustees Report.

This means that, for at least the past 30 years, policymakers have been told repeatedly that contributions from workers’ earnings were insufficient to finance the benefits they were accruing. As a result, costs that should have been borne by workers over the past quarter-century have been left to future generations. Below, I have used estimates developed by the system’s actuaries and presented in various annual Trustees Reports to calculate the magnitude of the cost shift.

At that time, the official financing shortfall estimate was 2.13 percent of covered payroll—the portion of wages and salaries that is taxed under the legal formula—over the next 75 years. The tax that finances Social Security retirement, survivor and disability benefits at the time was 12.4 percent of covered payroll, split evenly between employers and workers. So, if the system had been adjusted fully in 1994 by increasing the payroll tax, it would have risen to 14.53 percent of covered payroll. Alternatively, benefits could have been adjusted on a prospective basis.

For my assessment of the generational effects of the failure to act back then, I have assumed the fix would have been accomplished by adjusting the tax, only because this simplifies the computations and because the policy change in either event would almost certainly have had no effect on those already retired by 1994. So, either way, the burden would have fallen on current and future workers.

For those born in 2012, who would be just entering the workforce in 2034, the gap-closing 2034 reform will cost them nearly two years more of their earnings than their grandparents’ generations paid relative to current law.

If the payroll levy had been raised to 14.53 percent in 1994, the added tax revenues and estimated interest income earned would have left the trust funds with $7.3 trillion in assets by the end of 2018. But the payroll tax was not increased, so the figure in late 2018 was only $2.8 trillion. Using trustee projections about future payrolls subject to the tax and future interest rates for accruing income, the trust funds would then have held $12.9 trillion at the end of 2033, and the system would be in positive balance through at least 2095—the end point of 75-year projections in their 2018 report.

The 2019 Trustees Report, by contrast, suggests that without the extra tax revenue the cupboard will be bare in 2035. And if adjustments to financing are delayed until the trust funds are on the edge of depletion, the projected tax needed to bring the system back into actuarial balance will be around 4 percent of covered payroll.

So, what does this all mean? Virtually no one sitting at their kitchen table talks about Social Security or anything else in terms of trillions of dollars. They seldom have the data, the technical tools or the motivation to assess one approach versus another on Social Security financing. Elected officials who talk to their constituents about Social Security hardly ever raise the issue of how one policy or another might affect Generation A versus Generation B. In part that’s because, when promising one group a benefit increase, it is inconvenient to explain that doing so means a cost increase for another. And, in part, it’s because the information needed to explain the trade-offs in relatively simple terms is not available to them.

But the Social Security actuaries do generate a host of estimates that, along with plausible assumptions, allowed me to approximate the consequences of past failure to bring the system into close actuarial balance and to lay out the alternative paths forward. The results of the calculations are presented in the table below. Column 2 shows the cost of participation (measured in years of income) in the system in terms of estimated lifetime earnings for workers under past and existing law.

Thus, the entry for the 1946 birth group—the leading edge of the baby boomers—estimates the cost at 5.10 years of earnings; the figure for the 1956 birth cohort was 5.35 years of earnings; and so forth. The portion of earnings for those born later is greater than for the earlier groups because the payroll tax rates have been raised periodically. The combined rate (worker plus employer tax) did not reach 12.4 percent (the current rate) until the 1976 birth cohort came into the workforce.

Column 3 shows the accumulated costs workers would have borne as a percentage of lifetime earnings if the combined payroll tax had been raised from 12.4 percent to 14.53 percent in 1994—the hike needed going forward from that year to bring the system into actuarial balance. The differences between columns 2 and 3, shown in column 4, are revealing. The later a birth cohort joined the workforce, the bigger the bite of reform as measured by years of estimated lifetime earnings paid into the system. Each birth group has gained by the delay in reform, but there’s no getting around the fact that the delay will raise costs for others down the road, while possibly letting others off the hook for paying the price of required fixes.

Now fast-forward. Column 5 shows the costs of participating in Social Security if reform is delayed until 2034, when the combined payroll tax will need to be raised by 4 percentage points, amounting to 16.4 percent of covered earnings. Note in column 6 that if reform is put off until 2034, the baby boom generation is going to escape scot-free from helping to pay for the Social Security financing that loomed over their working careers. By contrast, for those born in 2012, who would be just entering the workforce in 2034, the gap-closing 2034 reform will cost them nearly two years more of their earnings than their grandparents’ generations paid relative to current law.

Column 7 shows the cost of not fixing the shortfall in 1994 versus the cost of waiting until 2034 to address the problem. For the oldest members of the baby boom, the net savings will be about one-quarter of a year of their lifetime earnings on average. For those born in 2012, the net cost shift of delaying reforms will be around 80 percent of a year’s worth of lifetime earnings.

Who Should Pay the Cost of Delay

Polls asking the public whether they favor higher payroll taxes to address Social Security’s financing shortfalls or reductions in benefits typically tilt strongly toward tax increases. When we consider those results, we ought to take the numbers in the table into consideration. Polls might accurately represent the opinions of the adult population, or even of the working-age population, but have almost no representation of the very young, who will in fact foot the bill for the policies being advocated. Why should my (baby-boom) generation be given almost a free ride compared to workers just starting out?

Deferring to what he sees as public opinion, Gale suggests that, as a practical matter, workers over 55 at the time of reform can’t be asked to pay. He thus points to the urgency of tackling reform soon. If we delay reform another five years, the 55-and-older clause will encompass workers who are now 50 and over; 10 years gets us to 45 and over.

Now let’s put all this in political context. The 1966 birth group will be 55 in 2021; the 1976 birth group crosses into the proposed free-ride zone in 2031. Public opinion will probably continue to favor tax increases as the solution. But among most current workers, further delay will be to their benefit, given the reform framework they are being presented.

Why take anyone 55 and older out of the picture when we consider our options? The political clout of the elderly is without parallel; their economic interests are more focused, and they are more likely to vote.

My question: why take anyone 55 and older — even current retirees — out of the picture when we consider our options? Yes, the political clout of the elderly is without parallel; their economic interests are more focused, and they are more likely to vote. But there’s also a matter of equity.

The arguments for protecting current retirees and those near retirement are typically based on two premises. First, they have already paid for their benefits while working, and it is unfair to change the rules of the game just as they’re claiming their prize. Second, they generally have reduced incomes in retirement compared to their working years—and have less flexibility to adapt to unwelcome financial surprises than their working-age counterparts.

Well, not quite. The vast majority of older workers certainly did pay many decades’ worth of Social Security taxes on their earnings. But, by the same token, the accumulating Social Security financing shortfalls over the past three decades is evidence that the vast majority of those in the system today who are older than 50 will not have paid the full costs of the benefits they are owed during their working lives.

Moreover, the notion that the elderly have lower incomes and less flexibility to deal with policies that affect their disposable incomes than their working-age counterparts doesn’t bear up to close analysis. Since the mid-1970s, the Social Security Administration has published a series of reports on the income status of the elderly in the United States based on big surveys. And these reports have been the basis for views about the income status of the elderly over the years.

In 1995, I raised a concern that the surveys were reporting significantly less retirement income than federal income tax and other federal reporting indicated retirees were actually receiving. And my own subsequent research suggested this difference was growing over time.


Two years ago, Adam Bee and Joshua Mitchell, analysts at the Census Bureau, released the most comprehensive review of elderly income reporting that has ever been developed. They matched survey retirees’ responses about income in 2012 to individual federal income tax forms and Social Security benefit administrative records that focused on elderly income. They found that the incomes of the elderly were significantly higher than the official government measure of their income and poverty status suggested, that incomes were higher across the whole distribution by significant amounts, and that a large part of the income missed by the government survey was from retirees’ retirement pensions and savings plans.

All told, the Census’ Current Population Survey missed 21 percent of the elderly’s income. By comparison, the CPS accounted for all of the wage and salary incomes of 18-to-64-year-old survey respondents, when compared to income reported on their W-2 and 1099 income tax forms.

Moreover, using the administrative data for the 1990-2012 period, the researchers found that the median household income of those age 65 and older rose by 29 percent after adjusting for inflation — figures that held up across the whole middle range of the income distribution. By comparison, the CPS data indicated that the median income of working-age families with at least one 18-to-64-year-old full-time, full-year worker increased by a mere 2.3 percent from 1990 to 2012 in inflation-adjusted dollars, declined by 4.6 percent at the 25th percentile of the distribution and increased by 9.3 percent at the 75th percentile. Thus, income growth for the elderly during this period considerably outstripped that of working-age families.

A separate analysis by Peter J. Brady, Steven Bass, Jessica Holland and Kevin Pierce at the IRS used a sample of individuals’ federal income tax returns running from 1999 through 2010 to examine the economic transition of older workers to retirement. The sample consisted of individuals who were ages 55 to 61 in 1999 and employed. The goal of the analysis was to assess how workers’ income changed from the year prior to their claiming Social Security to the year they claimed it and the subsequent three years.

I have argued for years that we should consider raising the Social Security benefits of lower earners, because I believe that retirees who have spent a full career at low wages deserve pensions that keep them out of poverty.

Now, for years, economists have puzzled over the decline in income that was frequently observed in surveys covering this transition. In fact, the new analysis suggests much less of a drop-off than previously found. For the lowest three-fifths of the distribution, median income two years after claiming Social Security benefits was greater than in the year before claiming benefits. There was some fall-off in income in the upper half of the elderly income distribution. But remember, the aforementioned Census Bureau analysis shows that income for the elderly at the 75th percentile of the income distribution has been growing much more rapidly than for any part of the distribution for working-age families.

What Next?

In January 2019, John B. Larson (D-CT), chair of the House Ways and Means Committee, introduced the Social Security 2100 Act with 200 co-sponsors. (The bill has also been introduced in the Senate.) Among other things, it calls for benefit increases above the current mandate for all current and future beneficiaries. Financing would change in two ways. First, the payroll tax would gradually increase until the combined employer-worker levy rose by 2.8 percentage points over the current 12.4 percent rate. Second, all earnings over $400,000 per year would be taxed. There’s no fiscal legerdemain here: the Social Security actuaries concluded that, if implemented, it would resolve the long-term financing shortfalls that exist under current law.

I recently met with one of the Senate sponsors to explore the philosophical underpinnings of this proposed reform. I asked him to explain the basis for sharply raising the payroll tax on young workers in order to pay higher benefits to baby boomers, who had not paid the full cost of benefits they were already entitled to. His response: to make the new tax on $400,000-plus earners politically palatable, it needed to be coupled with higher benefits for higher earners. Yet hardly any current retirees would actually pay the added taxes, so why the need to provide the windfall?


I noted that it would take the payroll tax collections from 40 full-time workers in 2019 to finance the combined benefits that my wife and I would receive from the program this year. And I pointed to the evidence cited earlier that low earners have lost ground in recent years while retirees have enjoyed rising incomes.

I have argued for years that we should consider raising the Social Security benefits of lower earners because I believe that retirees who have spent a full career at low wages deserve pensions that keep them out of poverty. But to give them more, we should not simply take more from current workers. We should also ask higher-level beneficiaries to help meet the cost of bolstering the benefits for retirees who are heavily dependent on their Social Security pensions.

This could be accomplished in various ways. Among them: capping annual cost-of-living increases for pensioners, taxing all (rather than part of) Social Security benefits as regular income and limiting spousal benefits.

Indeed, I think it would make sense to eliminate the spousal benefit altogether, replacing it with a true joint-and-survivor benefit. The spousal benefit made sense in 1940 when just 1 percent of the retiring workers were women who earned their own benefits on the job. But that rationale hardly works today, especially in light of the fact that the spousal benefit tends to offset the progressive redistributive nature of the benefit formula that links pension amounts to past earnings. Substituting a joint-and-survivor benefit would far more efficiently reduce poverty among surviving spouses at lower cost.

The private pension decline is a subject for another day. But I will say that, in my view, it makes little sense to use increases in Social Security, which involve scattershot transfers of income between generations and income classes, as a substitute for reforms in private retirement savings.

Promises of heftier benefits may be popular with the elderly lobbies and their constituents. Those just starting careers aren’t paying attention, so the resistance will be tepid to nonexistent. But the judgement of history on policymakers who ask the young to fatten the retirement of the old won’t be pretty.

main topic: Pensions