Measuring Social Security Proposals
Measuring Social Security Proposals by More than Solvency: Impacts on Poverty, Progressivity, Horizontal Equity, and Work Incentives
View Original Article – Center for Retirement Research
Although a number of measures are commonly used to evaluate Social Security proposals, they are often somewhat limited. Actuarial and microsimulation models routinely show actuarial balance, average or median benefits, replacement rates at retirement, and ratios of lifetime benefits to lifetime taxes. Identifying who wins and loses under alternative policies at a point in time – for example by lifetime earnings quintile – has also become commonplace in dynamic microsimulation analyses of changes to the Old-Age, Survivors, and Disability Insurance program (OASDI).
Often lacking, however, is a more complete, nuanced discussion of how Social Security benefits change relative to targeted objectives, such as enhancing adequacy for those with the greatest economic need, reducing the unequal treatment of dual- as opposed to single-earner couples with the same earnings, and improving work incentives for dually entitled and longer-term workers. For example, some studies that primarily address Social Security adequacy or equity issues show winners and losers from a benefit change, without providing a good sense of how much need is reduced or whether losers include those the system currently favors. Do we care, for instance, about whether women gain, relative to men, or is the issue whether lower-income individuals, who might disproportionately be women, gain? Similarly, should we care about whether a proposal increases the equality of treatment of those in equal circumstances – a change that may increase overall shares winning and losing?
When it comes to objectives, public finance principles suggest that programs generally should be target efficient. Within a Social Security context, this generally means that benefits should be distributed in an actuarially fair manner, except when some specific target such as progressivity is sought.
2 Some efforts are more administrable than others, and some create problematic incentives, further complicating matters.
3 Distributional analyses of Social Security are extremely complex, partly because of the ways the program combines mandated saving and redistribution within and across cohorts (Burkhauser and Warlick 1981). While understanding how benefits will change at a point in time under a particular proposal is useful, placing these changes in the lifetime context is also important. Most workers pay Social Security payroll taxes for several decades, and then collect benefits for many years. While in retirement, they may pay personal income taxes on their benefits if their incomes exceed designated thresholds. Comprehensive analyses need to account for both the tax and benefit sides of Social Security. The fact that the U.S. Social Security program is currently underfunded exacerbates this complexity. Frequently analysts compare systems in equal balance (in budget parlance, they compare deficit neutral systems), but what those balanced alternatives should be, especially over decades, is unclear.
4 This paper explores and develops a series of measures that we then apply to simulation output from the Urban Institute’s Dynamic Simulation of Income Model (DYNASIM) under current law and an alternative, the proposal of the National Commission on Fiscal Responsibility and Reform (henceforth NCFRR, commonly known as the Simpson-Bowles Commission). Using these two laws for illustration, we show how some provisions aimed usually at redistributing benefits have spillover effects into some aspects of horizontal equity.
5 We also document how sensitive conclusions about distributional impacts are to analysts’ choices about how to compute different measures (for example, by comparing individual and household, or “shared,” measures and comparing projections with and without the NCFRR proposal’s half benefit). Our paper is organized as follows. We begin by describing Social Security law and the NCFRR proposal. Next, we describe our approach and model. We then examine the literature on Social Security outcome measures, first identifying complex and controversial issues. We proceed to catalogue commonly used individual- and family-level measures of adequacy, progressivity, horizontal equity, and work incentives from other literature, discussing their respective strengths and weaknesses and relationship to principles behind Social Security and criteria for reform (e.g., Steuerle and Bakija 1994, Walker 1999).
6 We consider not just simple parameter measures of outcomes, but show outcomes for various classes of the population determined by lifetime earnings, taxes paid, and other characteristics. Exhibits 1 through 8 contain the heart of this paper: accessible measures that can be produced as part of a standard set of outputs by which to compare different OASDI proposals. In addition to discussing the measures briefly, we apply them to the NCFRR proposal as compared to current law. Finally, we close with some thoughts about how one could modestly restructure distributional analyses to raise the profile of issues beyond financial balance and attaining program solvency. Appendices discuss additional issues. For example, Appendices I and III discuss technical matters that are important for distributional analysis, while Appendix II describes the measures that the major U.S. government forecasting agencies, most notably the Social Security Administration (SSA) and Congressional Budget Office (CBO), use in their analyses.
