Written by: Eugene Steuerle
In a previous column, I raised the core question that Congress and the President cannot avoid, even if only indirectly, when undertaking Social Security, Medicare, and broad fiscal reform. The question: how much should the young transfer to the old? The answer determines the structural foundation of our federal government’s dominant social policy and a majority of its domestic outlays—that is, outlays other than defense, international affairs, and interest. The choices made in previous Social Security and Medicare reforms in 1935, 1938, 1965, 1972, 1977, and 1983 have had profound effects—in my view, largely but not always for the good—for decades and, in some cases, close to a century. Regardless, the choices Congress will soon make will continue to dramatically affect the economy, politics, and how future generations will structure their lives of work and leisure across their lifespans.
Past Social Security reformers have long considered this issue when proposing to “replace” a portion of a typical older person’s income with reference to what a typical younger person earns (more on this below). Unfortunately, that approach is now less wrong than incomplete. In deriving their formulas for what they called “replacement rates,” those reformers failed to account for multiple layers of transfers from workers to beneficiaries and for the effect of taxes on workers’ net incomes after those transfers.
Here, I want to lay out how to address, more comprehensively and explicitly, the overriding question of what both the old and the young would expect under any reform. This note will get a bit technical and wonky. The main issue I want to convey, however, is simple. Future reformers should retain the notion that desirable outcomes for the old in each generation should relate to the level of well-being attained by the young. However, for a variety of reasons—including the declining support that the federal government has been providing to the young and the working class—future reformers can no longer ignore that this is a two-way street. They must also address how the well-being of the young relates to that of the old.
Social Security’s Long-Time Answer to the Appropriate Level of Transfer
Social Security has long provided a partial answer to the question of how much to transfer from the young to the old. Since the 1970s, Congress has set the benefit for an average person retiring at the “full” retirement age (today, 67; for most of the past, 65) at approximately 40 percent of the economy’s “average” wage. While that rate is typically called a “replacement rate,” the formula more precisely relates a typical new beneficiary’s benefit at full retirement age to a measure of average earnings in the economy. This Social Security design allows the annual benefit for a typical member of a new cohort of retirees to be “wage-indexed,” so it grows from cohort to cohort at roughly the same rate as the average wage in the economy. Thus, if average wages grow 30 percent in inflation-adjusted dollars from one generation to the next, so will the average annual benefit.
There are many bells and whistles to this calculation. For example, for workers with lower lifetime earnings, the rate increases, while for those with above-average lifetime earnings, it decreases. See the chart below. Among the many items not shown in the chart, the rate increases on average for workers who have spouses and survivors; the lifetime replacement rate grows faster than the annual replacement rate because younger cohorts receive more years of benefits than older ones; and the “wage” is measured based on cash income and excludes other forms of compensation.
Source: Copied from Timothy Taylor, When Social Security Went Haywire, based on Andrew Biggs and the Social Security Trustees Report.
Adjusting the Effective Replacement Rate for Taxes Paid and Number of Taxpayers
What I want to raise here is how to use the same type of relative comparison when setting objectives for a future reform, while accounting for two previously ignored factors. First, the tax rate required of the young increases when their numbers decline relative to the number of people the government designates as eligible for “old age” (as well as disability) support. This factor, called the worker-to-beneficiary ratio, declines both when the birth rate falls and when the government provides more years of support as people live longer.
Consider a society that once had four workers for every person receiving support. If the level of support averages 40 percent of the average wage, then in a pure, flat-rate, pay-as-you-go payroll tax system like Social Security, each worker would need to contribute 10 percent of their earnings to support each beneficiary. That number appears on the left side of the lowest line on the chart at the top of this essay.
Note that the beneficiary’s income is not simply 40 percent ($40 for every $100) of the worker’s earnings, but 44 percent ($40 out of $90) of each worker’s after-payroll-tax earnings. Let’s call the 44 percent rate the “effective replacement rate.” The difference between 40 percent and 44 percent might not seem large, as it did in Social Security’s early years. But stick with me.
Consider next how our society continually increases the number of beneficiaries relative to the number of taxpayers. In the United States, people live longer, retire earlier, and a larger share of each cohort is likely to live until retirement eligibility—all of which has to be paid for somehow. If people retired with the same remaining life expectancy as in 1940, when Social Security first paid benefits, they would retire at about age 78 rather than at about 65 today.
Meanwhile, the fertility rate has fallen sharply, further reducing the number of workers available to support each beneficiary.
As a result of these demographic shifts, about one-third of the adult population now becomes eligible for Social Security benefits for about one-third of their adult lives. And these fractions keep increasing.
Most projections, therefore, show Social Security’s worker-to-beneficiary ratio falling from about 4-to-1 in the early 1960s to 2.7-to-1 today, 2.3-to-1 in about 20 years, and closer to 2-to-1 within another few decades.
If the desired replacement rate remains at 40 percent in a simple pay-as-you-go payroll tax system, each worker would now need to contribute $20 for every $100 earned. (This can again be seen in the chart at the top by holding the replacement rate at 40 percent and then moving from the 4-1 to the 2-to-1 worker-to-beneficiary ratio.) As a consequence, the worker’s after-tax income would fall to $80, and the effective replacement rate would be 50 percent.
Now consider what happens with other programs, such as Medicare and Medicaid long-term care, stacked on top of Social Security. These programs are also significantly financed by burdens on workers, whether through the Social Security tax, other taxes, or deficits left to future taxpayers. Suppose this also entails an additional 40 percent “replacement rate.”
Then, if two workers again support each beneficiary, each worker on average must contribute 40 percent of their earnings to achieve an 80 percent subsidy rate for the retiree relative to the worker’s before-tax wages. (See the top line in the top chart and move to higher replacement rates.) Since that tax rate leaves the worker with only 60 percent of earnings, the effective income of a typical older person in this simple example now rises to 133 percent of the average worker’s wage.
Now, there are several ways both the current and a reformed system might address these issues. It can raise taxes on older people or on capital income, not just on workers’ earnings. It can adjust the retirement age so that people don’t continually receive more years of support as they age. It can do a better job of encouraging people to stay in the workforce, thereby raising tax revenues. It can increase tax rates or expand the tax base for workers in various ways. It can also run larger deficits, a strategy employed especially by recent presidents and Congresses. Of course, these deficits only add to the interest costs borne mainly by tomorrow’s workers, further exacerbating the longer-term problem. And, to be clear, most of these adjustments would still involve transferring from the young to the old.
Attempting to make each of these adjustments fairly and efficiently would require more elaborate calculations and discussions than I provide here. However, if reformers follow past practices and try to stack up spending and tax reforms one at a time in each program to hit separate budgetary targets, they will almost assuredly fail to answer the explicit question of how much they require the young to provide to the old.
Including the Tax Rate in the Calculation of an Effective Replacement Rate
I promised up front that I would show a way to address more rigorously the question of how much the young should provide to the old. I am not proposing an idealized size for that transfer, but simply a way—as the original notion of a “replacement rate” did—to consider the choices.
What I have demonstrated so far is that it makes no sense to calculate a replacement rate independently of the taxes required to fund it and of the income and well-being of the workers paying those taxes.
Suppose that reformers decide that the replacement rate—or, more precisely, the benefits for a typical retiree—should equal 80 percent of an average-income worker’s after-tax income, not 80 percent of that worker’s earnings. Then the required tax rate per worker, when the worker-to-beneficiary ratio is 2-to-1, would be 29 percent, not 40 percent. In that case, the beneficiary would receive the equivalent of 58 percent of a worker’s after-tax earnings, and each worker would have left 71 percent of their earnings (that is, 58% for a retiree, divided by 71% remaining for a worker, equals 80 percent).
The graph below compares the tax rate required to achieve an effective replacement rate, which accounts for taxes paid by workers, with a replacement rate that doesn’t account for taxes paid by workers. It does this for various replacement rates when the worker-to-beneficiary ratio falls to 2-to-1.
Do you think even a 29 percent tax rate is too high? If so, consider other reforms. For instance, suppose reformers remove many of the ways that Social Security discourages work, so the worker-to-beneficiary ratio stabilizes at 2.5. The bottom line in the chart above would fall further, and the required tax rate at an effective replacement rate of 80 percent would be about 24 percent.
As noted, I am not trying to specify which reforms to undertake. I am merely claiming that reformers should set targets based largely on the needs and abilities of the entire population over time. To the best of my knowledge, no recent Social Security reform proposal designed to date does that. To be honest with the public—and to help justify why reform must address unsustainable current policies, reformers should measure and make explicit how much more they think we should — and can — continue to provide from the young to the old under different future economic and demographic conditions. One way to do that is to take into account the tax rate required for all transfer programs when calculating “effective replacement rates” for people Social Security defines as “old” and in need of support. The existing measure of replacement rate cannot suffice. More broadly, the Social Security Administration and the Congressional Budget Office should include in their assessment of any proposed or adopted reform a comprehensive comparison of effects on the well-being of young and old alike.
Note: Among my forthcoming notes will be an examination of how, largely due to the factors discussed here, the young are already more likely than the old to fall into lower-income classes. I will also show how close some of the hypothetical numbers shown here are to the real world.
Related: Clients Don’t Start With You Anymore—Here’s What That Changes