Summary and Introduction
The Social Security benefit formula has incorporated some measure of progressivity almost since the program's inception. As early as 1939, amendments to the original Social Security Act stipulated that monthly benefits replace a higher proportion of preretirement earnings for people with lower earnings compared with those with higher earnings (Martin and Weaver 2005). Throughout the program's history, benefit adequacy—promoted through a progressive benefit formula—has been balanced with equity, the goal that benefits increase with contributions.
Although the Social Security retirement system is designed to replace a higher percentage of earnings for lower-income workers and their dependents,1 the degree to which the program actually achieves progressive outcomes is less certain. To contribute to our understanding of this issue, this paper introduces a new measure—a progressivity index—to estimate the progressivity of the Social Security retirement program. Our purpose is to develop a comprehensive, easy-to-understand summary measure for evaluating the magnitude of progressivity under Social Security within cohorts and for assessing potential shifts in systemwide progressivity as a result of policy changes.
Social Security progressivity can be described in various ways. For the purposes of this paper, we define progressivity as the degree to which benefits are higher relative to lifetime payroll contributions for lower contributors than for higher contributors. This definition can be related to Musgrave and Thin's note (194
regarding the income tax:
It is generally agreed that a rate structure is progressive where the average rate of tax (i.e., tax liability as a percentage of income) rises when moving up the income scale; proportional where the average rate remains constant; and regressive where the average rate falls with the rising income.
Progressivity under the Social Security program, which both levies taxes and provides benefits, is necessarily more complex than for income taxes. While this paper's definition of progressivity is just one approach, and others have merit, the definition used here appears to be consistent with Social Security's program design.2
The progressivity index compares the distribution of the present value of lifetime benefits to the distribution of the present value of lifetime taxes. We apply this index to microsimulation data from the Social Security Administration's (SSA's) Modeling Income in the Near Term (MINT) model and estimate progressivity under the program for those born between 1926 and 2017.3 Results indicate that Social Security is modestly progressive on a lifetime basis; currently, the program lies approximately halfway between paying a benefit directly proportional to lifetime taxable earnings and paying a flat dollar benefit to each retiree. Although the program's progressivity has declined in recent decades, it is projected to remain roughly constant in the future, according to the index. Overall, the paper extends previous research by introducing a comprehensive, easy-to-understand summary measure that can evaluate Social Security's progressivity among current and future retirees or as a result of policy changes.
The paper begins with a review of the methodological techniques often used to measure progressivity under the Social Security program. We then introduce the progressivity index, describing how the index is calculated and how results can be interpreted. Next, the progressivity index is applied to MINT data of the Social Security population. The magnitude of progressivity under the current Social Security program is estimated and compared with two stylized hypothetical programs with high and low progressivity. This is followed by an examination of progressivity for future retirees. To demonstrate the index's potential utility in evaluating policy changes, the final section calculates the effects of several commonly cited Social Security policy changes on the system's progressivity.
Measuring Progressivity under Social Security
Progressivity is generally conceptualized as a function of redistribution between different groups of individuals, for Social Security purposes generally within the same birth cohort.4 The program is portrayed as progressive when it redistributes resources from higher- to lower-earning groups. While much of the existing work in this area has found the program to be somewhat progressive on a lifetime basis (CBO 2006; Leimer 1999, 2003), the issue of whether low earners fare better under the program than high earners, and in what ways, remains an open question.
It is important to distinguish from the outset between progressivity and the reduction of income inequality (Gustman and Steinmeier 2000; Coronado, Fullerton, and Glass 2000). As noted above, progressivity is defined in this paper as the degree to which individuals with lower payroll contributions receive higher lifetime benefits relative to their lifetime contributions than do individuals with higher payroll contributions. Although progressivity will tend to reduce income inequality, the overall reduction of income inequality is also a function of the size of the program. Put another way, a small but highly progressive program may do little to reduce overall income inequality relative to a large but modestly progressive program. Herein, we focus solely on the progressivity of the benefits and taxes, thereby controlling for Social Security's size relative to overall income, and do not analyze the degree to which Social Security reduces overall income inequality.
A number of methods can be used to assess the degree to which the Social Security program is progressive, each with its own advantages and constraints. Many studies attempt to evaluate Social Security's progressivity using replacement rates or money's-worth measures. Replacement rates measure how much preretirement income is replaced by the initial Social Security benefit. Although useful for measuring benefit adequacy at retirement, replacement rates cannot account for differences in lifetime benefits; for example, socioeconomic and demographic factors can offset the program's progressive benefit formula. Early studies such as Freiden, Leimer, and Hoffman (1976) and Aaron (1977) called attention to the fact that differential mortality rates across earning levels may offset the progressivity indicated by the replacement rate of a person's initial Social Security benefit at retirement. In a more recent study using the representative worker approach, Beach and Davis (199
argued that African-Americans, while perhaps receiving higher replacement rates than whites due to comparatively low lifetime earnings, may actually receive lower returns from Social Security because of higher mortality rates (see also Garrett 1995 and Panis and Lillard 1996).5 Furthermore, myriad alternative replacement rate measures (Mitchell and Phillips 2006) can lead to confusion between studies.
By contrast, money's-worth measures take a lifetime approach by assessing the balance between lifetime benefits received against lifetime taxes paid under the program. A variety of approaches assess whether beneficiaries "get their money's worth" under Social Security, including the internal rate of return on accumulated contributions, the net present discounted value of taxes and benefits, and the ratio of the discounted present values of benefits to taxes.6 Overall, money's-worth measures tend to yield more informative results than replacement rates alone in terms of progressivity because they account for lifetime taxes paid and benefits received.
Note that both replacement rates and money's-worth measures do not measure progressivity on their own, but rather are measures whose results must be compared across indicators of economic well-being to determine the level of progressivity. One common strategy is to compare replacement rates of "stylized" or "hypothetical" workers of different earning profiles (such as low, medium or high earners). If replacement rates increase as income or earnings decline, the program is described as progressive. However, this stylized worker approach does not incorporate the full diversity of lifetime outcomes, so the results may not necessarily be considered representative of the population being studied.
A common technique of evaluating progressivity using money's-worth measures is to break out results for a representative population by income quintiles. If the average internal rate of return or the benefit/tax ratio is lower for high lifetime earners than for low lifetime earners, then the system can be portrayed as progressive. Studies such as Cohen, Steuerle, and Carasso (2001, 2002), which reported modestly progressive redistribution from higher to lower income groups, tend to follow this approach. Money's-worth measures are also often examined in relation to socio-demographic characteristics such as education, race, and sex. In this vein, Leimer (2003) uses Social Security administrative data to compare women's benefit/tax ratio and internal rates of return to those of men.7
Money's-worth measures are extremely useful in assessing the treatment of a given individual and certain groups by the Social Security program. Breaking out money's-worth results by various subgroups can help explain differential returns to Social Security. However, it can at times be difficult to determine if the program is progressive in general. Consider, for example, examining money's worth by income quintile. If the ratio of benefits to taxes substantially increases for the second lowest quintile but falls slightly for the lowest, it would remain unclear if overall progressivity increased or decreased. Furthermore, there can be large variations in money's-worth values within earnings quintiles, which would go undetected without a finer breakdown of results. This is illustrated by a policy change in which the top half of the lowest quintile experiences a gain in money's worth, while the bottom half experiences a loss. In such cases, which are not uncommon in practice, the impact on average progressivity would be ambiguous.
To address some of the ambiguities mentioned above, another approach would be to develop a measure that produces a single measure of progressivity. To this end, Coronado, Fullerton, and Glass (2000) calculate the ratio of after-tax to before-tax total income, indicating progressivity by the degree to which the ratio is greater than or less than 1. Such a measure adds another useful tool for policy analysts because it provides a single number for progressivity, which makes comparisons easier and offers a measure of the distribution of total dollars across the entire population, rather than comparing individual measures (a worker's rate of return, for example) for different population segments.
However, the Coronado, Fullerton, and Glass approach measures how Social Security affects the progressivity of total income ("external" progressivity) rather than the progressivity of benefits relative to taxes ("internal" progressivity), which is the focus of this paper. Although desirable for many research questions, the Coronado, Fullerton, and Glass approach allows the size of the benefits to drive much of the effect, potentially producing counterintuitive results. For example, a small, highly progressive benefit, such as Supplemental Security Income (SSI), may have a negligible effect on the progressivity of the income distribution. Although the level of average benefits is obviously a significant question for policymakers with important implications for public welfare, we choose here to measure the progressivity of benefits distinctly from the amount of benefits provided.
The Social Security Progressivity Index
This paper develops a new measure referred to as a progressivity index to complement existing measures of progressivity. This index provides a summary measure of the progressivity of taxes and benefits on a lifetime basis. The index could be an intuitive measure for policymakers because it essentially provides a thermometer-like gauge for progressivity.
The index's values generally range from 0 to 1, allowing for easier comparisons of progressivity on a systemwide basis. A value of 0 represents no redistribution: Lifetime benefits are exactly proportional to lifetime contributions. This can be equated with a pure defined contribution (DC) pension program without annuitization. A value of 1 represents an extreme scenario in which all taxes are paid by the highest-earning individual and all benefits received by the lowest-earning individual.8
The progressivity index is constructed using Lorenz curves adapted to show the distribution of lifetime Social Security tax payments and lifetime Social Security benefit receipts. Chart 1 illustrates such curves using stylized data. Individuals are sorted by their lifetime taxes, and the horizontal axis represents the cumulative percentage of total lifetime taxes paid while the vertical axis represents the cumulative percentage of total lifetime benefits received.9 Thus, a given point on the line indicates that individuals paying x percent of total lifetime taxes receive y percent of total lifetime benefits. A line with a slope of 1, labeled as the "no-redistribution line," represents a program in which lifetime benefits are precisely proportional to lifetime taxes. A curve above the no-redistribution line represents progressivity, as it shows that individuals or households paying a given percentage of total taxes receive more than that percentage of total benefits. A curve below the no-redistribution line represents regressivity.
Logged
