MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

January 2018 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD Expanding Qualifying Credit Options for Social Security Benefits Christian Weller, McCormack Graduate School of Policy and Global Studies, University of Massachusetts Boston, Boston, MA Darrick Hamilton, Milano School of International Affairs, Management and Urban Policy, and Department of Economics, New School for Social Research, The New School, New York, NY

This paper represents the views of the authors and does not necessarily reflect the views or policy of AARP or the opinions or policy of any agency of the federal government nor of any of the educational and research institutions that sponsor their work.

CONTENTS Introduction 1 Policy Innovations 3 Caregiving Credit 3 Credit while Unemployed 5 Credit during Job Training 5 Additional Implementation Considerations Related to New Ways to Earn Credits 6 Current Policy 6 Context of Similar Past Proposals 7 Potential Consequences of Our Proposal 8 Target Populations 10 Options to Pay for Additional Benefits 15 Conclusion 18 Appendix 19 References 20 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD i

TABLES TABLE 1. Share of People Age 62 and Older with Family Income Below 100 Percent of the Federal Poverty Line, by Demographic Characteristic for Selected Years with and without Policy Innovations 11 TABLE 2. Dollar and Percent Change in Average Per Capita Annuity Income of People Age 62 and Older by Selected Characteristics and Selected Years 12 TABLE 3. Percent Change in Per Capita Annuity Income among People Age 62 and Older by Selected Characteristics for 2065 14 TABLE 4. Change in Average Net Per Capita Cash Income of People Age 62 and Older by Selected Characteristics and Selected Years (in 2015 Dollars) 16 TABLE 5. Indicators of Fiscal Changes for Select Years 17 ii MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

INTRODUCTION Social Security has gained importance in a world of increasingly volatile incomes. Social Security benefits stabilize retirement incomes, especially for people who experience earnings fluctuations and unemployment spells during their careers. As a result, retirement incomes fluctuate less than people s earnings do during their careers. The need for such stabilization of retirement incomes has increased as earnings, work schedules, and expenses have become more volatile (Dynan, Elmendorf & Sichel, 2007; Ziliak, Hardy & Bollinger, 2011; Hardy & Ziliak, 2014; Hardy, 2017). Researchers have identified a number of key trends underlying the increase in income and expense volatility since the mid-1980s. These trends include more caregiving responsibilities due to aging, more single-headed households (Boushey, 2016; Ellis & Simmons, 2014), longer unemployment spells, and less labor market mobility with stagnant wages (Kopczuk, Saez & Song, 2010). Moreover, increased labor market uncertainties, such as the ones associated with caregiving responsibilities, potential unemployment spells, and unstable jobs, adversely hamper people s ability to save (Weller, 2015). This is more pronounced for individuals from marginalized or stigmatized groups based on race, ethnicity, and gender. Their jobs and earnings are more precarious and they tend to have more caregiving and financial responsibilities due to extended kin and friend networks of other people also living in poverty (Chiteji & Hamilton, 2002), thus contributing, for instance, to the persistent racial wealth gap (Hamilton & Darity, 2009, 2010). People need stronger Social Security retirement benefits to partially offset the increase of earnings and expense volatility. Workers could enjoy more stable benefits if they earned credits toward benefits for caregiving during periods of unemployment and when pursuing MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 1

job training programs. This would help people not lose all of their retirement benefits when they experience earnings shocks, even if their earnings disappear. People also need contemporaneous assistance to improve their economic security while they experience People also need income volatility. Our proposals are complements contemporaneous (not substitutes) to improve assistance to improve benefits such as paid their economic security family leave, stronger unemployment insurance while they experience benefits, and expanded income volatility. public-sector infrastructure jobs. 1 Policy makers should also consider pursuing other targeted Social Security benefit improvements, such as a more effective minimum benefit that could boost benefits for the lowestincome earners. Our proposals, while targeted and effective, should be seen as part of a larger agenda to improve the economic security of Americans in general, and vulnerable populations specifically. In our paper, we first detail proposals to update Social Security, including administrative specifics. Next, we briefly discuss the target audiences and their earnings experiences. We specifically show that women and communities of color could disproportionately benefit from our proposed changes due to disproportionate caregiving responsibilities, unemployment risks, and limited earnings mobility. We discuss a number of key simulation results that show the effect of our proposed changes on people s benefits. Our innovations would strengthen Social Security s income stabilization in a targeted manner at a time when the need for such income is growing. Further, we highlight the likely costs associated with our proposed changes and show that imposing some payroll taxes on capital income will not only help to pay for the proposed benefit changes, but also shrink Social Security s long-term deficit, thus stabilizing benefits for all beneficiaries. 1 Such additional contemporaneous benefits should be delivered through programs separate from Social Security with their own funding streams, to avoid a dilution of existing Social Security benefits. 2 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

POLICY INNOVATIONS We propose three changes to Social Security benefit calculations that directly counter some of the underlying causes of potential earnings declines. Such earnings declines could result from, among other things, the altruism of caregiving for family and friends as well as from increasing labor market volatility, such as longer spells of unemployment and less upward earnings mobility. All three of our proposed changes would affect the way people earn credit toward Social Security benefits. Social Security retirement benefits are ultimately based on people s average earnings for their 35 years with the highest earnings. 2 Most important for our purposes, this means that, if a worker had fewer than 35 years of earnings, the missing years are counted as years with zero earnings when computing Social Security benefits. These zero-earnings years then reduce average lifetime earnings and consequently Social Security benefits. People can have zero-earnings years, for instance, if they care for a family member, are unemployed, or take time away from paid work for job training to increase their future earnings potential. Their benefits end up being less than if they had continuously worked. CAREGIVING CREDIT First, we propose allowing workers to earn credits for caring for a family member or friend, starting in 2018. 3 A worker can earn credits toward Social Security benefits for caring for somebody else at least 20 hours per week. This includes children up to the age of 10 as well as ailing family members and friends. Caregivers could receive up to three years worth of credit. For a married couple, each spouse can qualify for the maximum years of credits, although only one spouse can receive the credit in a given time period. That is, married spouses cannot share caregiving responsibilities and both receive credits toward benefits. Caregivers would be credited time at the amount of 60 percent of the average 2 People need to earn a minimum number of credits for quarters to become eligible for benefits. The amount of benefits are then based on the average earnings during the time prior to receiving benefits. In 2016, a worker needed $1,260 of earnings to earn one quarter of credits. That is, a worker who earned at least $5,040 in 2016 received four quarters worth of credit for that year. A worker needs 40 quarterly credits to be eligible for retirement benefits, in addition to reaching age 62 to be eligible for at least early retirement benefits. 3 Workers could receive caregiving credit only for caregiving completed in 2018 or thereafter. There is thus a natural phase-in of this new benefit as people can earn up to three years of credit under our proposal. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 3

wage index (AWI), currently $49,121 for 2016, as long as they earned 60 percent or less of the AWI. Otherwise, they would receive credit between 60 percent and 105 percent of their earnings. Those earning at least the AWI in a given year would be credited 105 percent of their earnings. 4 Furthermore, credits would be prorated for the actual time of caregiving from one quarter up to three years. 5 Moreover, people who already receive Social Security retirement benefits would also see a benefit from caregiving. The Social Security Administration would recalculate their monthly benefits by replacing People who already past zero-earnings years with receive Social Security earnings equal to 60 percent retirement benefits of the AWI during that year 6 or, if somebody has no zeroearnings years, increase his or would also see a benefit from caregiving. her average indexed monthly earnings (AIME), which determines the amount of benefits somebody receives, by one-thirtieth, or 3.3 percent. 7 Either approach would permanently increase retirement benefits in annual steps, after caregiving is completed. People could hence see up to three increases in their retirement benefits if they did not get any caregiving credits prior to retiring and if they cared for somebody for three years while in retirement. The caregiving credit faces a unique challenge. There is typically no record for most people who perform unpaid or very low-pay caregiving work for a family member or friend. Receiving a caregiving credit, though, requires that the Social Security Administration receive some certification that people performed caregiving work. For parents caring for a child, it will suffice to document that the child lived with them. Grandparents caring for a child would need to either show that they are legal guardians or present a notarized statement from the parents that they are responsible for substantial care at least 20 hours a week for a grandchild. For other caregiving work, such as elder care or for caring for somebody with disabilities, caregivers could similarly collect verifiable documentation. This would include a letter from the person they cared for or his or her guardian certifying the caregiving services were performed. They would also collect documentation that the person needed care. 8 Such documentation could be a Medicaid eligibility determination for long-term and inhome care services, Social Security Disability 4 The appendix includes a detailed formula for the calculation of the caregiving credit. 5 The appendix includes hypothetical examples for this option and for two alternatives. 6 The Social Security Administration would first replace the year with the highest wage-adjusted substantial earnings if somebody had more than one year of zero earnings. 7 We choose one-thirtieth here even though benefits are calculated on the basis of the highest 35 years. The logic is that those with full career earnings 35 years should receive a relative increase in their benefit that resembles that of somebody who did not have a full career. 8 These documentation requirements are similar to those for California s Paid Family Leave. See http://www.edd.ca.gov/ disability/faq_pfl_benefits.htm. 4 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

Insurance benefits, private disability insurance benefits, and a physician s functional assessment of that person showing that the cared-for person cannot live independently without care. In all instances, caregivers would collect and retain the required paperwork similar to documentation for business expenses claimed on tax returns. To minimize the burden on caregivers, the tax authorities would include a short new line on tax forms showing that somebody was a caregiver for a specified amount of time during a particular tax year. That is, people would indicate that they were caregivers on their tax returns and retain the relevant documentations for a reasonable timeframe, typically three years. CREDIT WHILE UNEMPLOYED We also propose to grant workers credits toward Social Security benefits while receiving unemployment insurance (UI) benefits. Each unemployed worker could receive credit toward Social Security benefits based on Social Security s AWI specifically 60 percent of AWI and the length of time he or she were unemployed, up to six months for each eligible unemployment spell. To give an example, a worker who received UI benefits for three months would receive the same credit toward Social Security benefits as a worker who earned $7,368 one-fourth of 60 percent of $49,121 during this threemonth period. In 2016, this amount would have been sufficient to receive four quarters of credit toward Social Security benefits. A worker can receive up to three years worth of credit toward Social Security benefits while receiving UI benefits. 9 Unemployment insurance agencies will report names and length of unemployment to the Social Security Administration. CREDIT DURING JOB TRAINING Third, we propose to give workers who participate in unpaid or low-pay training programs such as Jobs Corp credit for their time. The dynamic nature of our economy requires continuous workforce retooling and training. Our proposal is meant to mitigate some of the risk and discouragement from these skill developments as it relates to retirement security. Workers would receive credit equal to substantial earnings $22,050 in 2016 for the time they spent in Our proposal is meant to mitigate some of the risk and discouragement from these skill developments as it relates to retirement security. 9 This potentially creates a wrinkle. Consider the following possibilities of two workers receiving unemployment insurance benefits for six months. The first worker receives benefits from March to September of one year, while the second worker receives benefits from September of one year to March of the next year. The first worker will receive four credits in one year under our proposal, while the second worker will receive four quarters of credit in one year and another four quarters of credit in the next year. This unevenness will not matter as workers will likely have enough earnings in each year to receive four quarters of credit in each year. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 5

training. Accredited job training programs will report those times to the Social Security Administration. Workers would have to be at least 21 years of age to receive such credits, under the assumption that training at earlier ages will likely consist of some college education. Workers can earn up to three years worth of credits for job training. ADDITIONAL IMPLEMENTATION CONSIDERATIONS RELATED TO NEW WAYS TO EARN CREDITS We briefly discuss a number of details related to the implementation here. First, workers could qualify for only one type of new credit at a time. That is, they could not claim two separate credits if they took on caregiving responsibilities while unemployed, for instance. Second, we propose that workers could earn a combined maximum of five years worth of credits for caregiving, unemployment, and job training during their careers. Specifically, they would receive credit for the five years of highest wage-adjusted earnings if they have Workers could qualify accumulated more than for only one type of new five years of such credits from the three new forms of credit at a time. earning credits caregiving, unemployment, and training. This would still require workers to have at least five years of paid work to qualify for their own retirement benefits thus maintaining the link between earnings and Social Security benefits. Credits above the five-year maximum will be recorded as zero or as actual earnings, in case people had some earnings from work in those years. The additional quarters above the five-year maximum will not count as quarters toward benefit eligibility. Third, our proposed changes automatically update each year as wages go up. The amount each worker is credited for caregiving during unemployment spells and training periods is tied to average wages. Because wages typically increase faster than inflation, our proposed credits will not fall behind workers productivity and living standards over time. Fourth, the credits would start in 2018. Workers will receive credits only for qualifying events caregiving, unemployment, and job training in 2018 and in later years. They will not receive credits for such events that occurred prior to 2018. This would mean that the full benefits gradually phase in over time. CURRENT POLICY Our proposal presents a meaningful improvement over existing policy by explicitly acknowledging the economic insecurities that workers face during their careers. Currently, workers do not receive credits for caregiving during unemployment and during job training. Workers can, however, receive an implicit benefit for caregiving through the spousal benefit. Spouses, typically women, who take time out from the labor market to care for 6 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

children, family members, and friends, but also due to unemployment and job training, may end up with relatively low benefits based on their own earnings. They will receive 50 percent of their husbands benefits if their own benefits are below that amount. The current spousal benefit, while important, falls short in a number of ways. It offers no benefit to single women and men, who increasingly take on child care and elder care responsibilities (Weller & Tolson, 2017). Similarly, it offers no benefit to single women and men during unemployment spells and when they pursue job training. It offers very limited income protections in a labor market where women increasingly have their own earnings records and men struggle with growing job and earnings instability. Our proposal thus is intended to offer greater income security to workers who face increasing labor market insecurities in a more targeted fashion than the existing spousal benefit does. CONTEXT OF SIMILAR PAST PROPOSALS Our proposal builds on a number of previous proposals, especially related to improving the economic security of caregivers, but also addresses challenges for unemployed workers (Favreault, 2008; Herd, 2009). These previous proposals follow two separate paths. One approach creates a new minimum benefit for those who have relatively low earnings over the course of their careers. The other approach would change the way workers can earn credits toward Social Security benefits. In those proposals, workers could earn additional credits or they could drop low- or zero-earnings years in the benefit calculation. The German equivalent to Social Security, Rentenversicherung, gives workers credits for child care during spells of unemployment and training (DRV, 2016). The German system shortens the minimum years necessary to receive benefits rather than crediting workers with specific amounts. In that sense, the proposal by Rep. Patrick Murphy (D-Fla.) for child care credits mirrors the German provision (SSA, 2016b). We see our proposal as a complement to creating a new minimum benefit for low lifetime earners, not as a supplement, as we discuss We see our proposal as a in a little more detail below. complement to creating Minimum benefits target very low lifetime earners, a new minimum benefit but they can come with for low lifetime earners, their own administrative not as a supplement. challenges (Herd, 2009) and they do not increase the income security of workers with modest earnings. Making it easier for workers to earn Social Security benefits at moderate levels instead addresses the labor market challenges that a wider range of workers face. The design challenge is to set the amount that is credited to the worker high enough to improve his or her benefits in a meaningful MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 7

way relative to current policy, but also avoid giving a disproportionate benefit to higherincome earners who may not need the extra assistance. We believe that our proposed credits meet this dual challenge. They are higher than previously estimated proposals (Favreault et al., 2006; Favreault, 2008; GAO, 2007) and expand the reasons for such credits to include unemployment and caregiving. On the other hand, our proposal avoids disproportionately benefiting higher-income earners, as allowing workers to drop low- or zero-earnings years would (Herd, 2006). Our proposal strikes a balance that advances the economic security of workers struggling under increasingly uncertain labor market conditions. POTENTIAL CONSEQUENCES OF OUR PROPOSAL We aim to strengthen workers income security with our proposed changes to the way workers earn Social Security credits. To this end, it is worth noting a few key aspects of our proposal. We have attempted, for instance, to avoid inequities and make the new benefits progressive so that the lowest lifetime earners, and those with the greatest earnings volatility, could receive the largest We aim to strengthen increase in benefits. Our workers income security proposed caregiving credit, for instance, is targeted with our proposed toward low-paid or unpaid changes. caregivers. Low-paid or unpaid caregivers would receive a bump in their Social Security benefits. This recognizes the valuable societal benefit that caregivers perform and offsets some of their low or nonexistent pay. Caregivers who receive market pay above the threshold 60 percent of the AWI would receive a progressively smaller benefit under our proposal. Few caregivers who care for somebody else for more than 20 hours per week will have earnings at or above $29,473 (60 percent of $49,121). Similarly, our credit for times of unemployment favors lower-income earners over higherincome ones. Low lifetime earners will receive a higher benefit than higher lifetime earners. We cap the credit amount, mirroring caps on weekly unemployment insurance benefits. It is thus a small and limited offset for the fact that the eligibility criteria for unemployment insurance receipt favor higher-income earners with more stable jobs. Our job training credit would help all workers in job training programs. Most of these programs focus on lower-income earners, as we discuss below, thus adding to the progressiveness of a flat credit amount for job training purposes. It is possible, though, that our caregiving credit may give caregiving spouses in high-income, single-earner couples a disproportionately large benefit, making our proposal somewhat less progressive. This would be the case if receiving the new credit would mean that a spouse would qualify for his or her own benefit when that would otherwise not be the case. This possibility is likely very low. Workers would have to have at least seven years of qualified earnings in addition to the 8 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

maximum caregiving credit to qualify for their own benefit. Someone who stays out of the labor force is thus unlikely to earn his or her own benefit due to our new caregiving credit. Our proposed credits could interact with other current or proposed features of Social Security benefits. Allowing people to earn additional credits will raise their primary insurance amount alongside their AIME. That is, the additional credits will boost all types of benefits. This is intentional as workers need more protections against income fluctuations amid increasingly uncertain labor markets. Our proposal would also lower the number of workers who would have to rely on a proposed minimum benefit, which we consider an important complement to our proposed changes. 10 Similarly, our proposed credits could make it easier for spouses with limited earnings records to qualify for their own benefits. This would be especially the case if spousal benefits were reduced, which is counter to the goal of gender equity, and a proposal that we do not endorse. Our proposed changes could also have positive social spillover effects, especially through the creation of a caregiving credit. This would strengthen the economics of caregiving, formalize informal relationships, and provide dignity to this socially necessary work. Our proposed credit would specifically apply to all caregivers, whether paid or unpaid and whether they care for children, other family members, or friends. Such a caregiving credit would thus address the realities of the increasing need for caregiving in an aging society with a growing share of single-headed households and adults without children. Our proposed changes to the way people would earn credits toward Social Security benefits could have unintended consequences, specifically a reduction in labor supply. People theoretically could choose to wait a little longer to return to paid work, while earning additional credits toward Social Security benefits. Such a labor supply effect likely will be small, as our proposal provides only additional benefits in the future, not extra pay in the present, with one small caveat. Some workers, in theory, could decide to collect unemployment insurance longer than they otherwise would have to earn Our proposed changes to extra credits toward Social Social Security benefits Security benefits. Any such effects, if they existed, would would strengthen the be limited by the eligibility overall protections for criteria and time limits of those workers who need unemployment insurance. And, in the case of caregiving them most. and job training, where there may be fewer external limits on time, a lower labor supply more time caring for others and spent on training may actually be desirable. Overall, our proposal would strengthen benefits for low lifetime earners and reduce the costs of other proposals. In sum, our proposed changes to Social Security benefits would strengthen the overall protections for those workers who need them most. 10 See, for instance, SSA (2016b). MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 9

TARGET POPULATIONS The possibility of earnings declines disproportionately affects women (Fetterolf & Rudman, 2014; Johnson & Lo Sasso, 2006; Maume, 2008) and communities of color (BLS, 2016a, 2016b; Ellis & Simmons, 2014; Fryer, Pager & Spenkuch, 2011; Garber, 2016; Hamilton, Austin & Darity, 2011) due to caregiving, unemployment, and limited labor market mobility, all of which follows from occupational segregation by gender. Latinos and African Americans, for example, regularly have higher unemployment Earning additional rates than Whites, and their likelihood of receiving UI is credits will increase the roughly proportional to their benefits in a targeted share among the unemployed way for those most (Michaelides & Mueser, 2012). All available evidence affected by potential suggests that many of these earnings losses. factors have increased over time (Michaelides & Mueser, 2012; Weller & Tolson, 2017), contributing to rising earnings volatility (Dynan, Elmendorf & Sichel, 2007). Earning additional credits will increase the benefits in a targeted way for those most affected by potential earnings losses. On average, benefit increases from the additional credits tend to be small and are gradually phased in over long periods of time, as one would expect when it becomes easier for people to earn credits during their careers. Table 1 shows the change in the poverty rate among people ages 62 and older in the coming decades. The poverty rate starts to decline by 2055, relative to what it would have been without our proposed changes, and the overall decrease in the poverty rate is small. More important is that the simulations show the poverty rate drops more for our target audiences. Women, people who were never married, communities of color, and those with less education see their poverty rates drop more than is the case for men, married people, Whites, and those with more education (table 1). For instance, the poverty rate for women is 0.2 percentage point lower in 2065 than it otherwise would be, the poverty rate for people who were never married is 0.4 percentage point lower, the poverty rate for African Americans is 0.4 percentage point lower, and the poverty rate for those with less than high school degrees is 0.6 percentage point lower. These decreases in poverty compare to 0.1 percentage point declines for men, married people, and Whites and no change for those with college degrees (table 1). 10 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

TABLE 1. Share of People Age 62 and Older with Family Income Below 100 Percent of the Federal Poverty Line, by Demographic Characteristic for Selected Years with and without Policy Innovations All Year Current Law With Innovations 2025 2035 2045 2055 2065 2025 2035 2045 2055 2065 7.9% 7.4% 7.4% 6.7% 5.7% 7.9% 7.4% 7.4% 6.6% 5.5% Women 8.3% 7.7% 7.5% 6.7% 5.4% 8.3% 7.7% 7.4% 6.6% 5.2% Men 7.4% 7.1% 7.4% 6.7% 5.9% 7.4% 7.1% 7.3% 6.6% 5.8% Less than h.s. 22.4% 22.4% 22.6% 21.0% 18.7% 22.4% 22.4% 22.4% 20.6% 18.1% High school 8.7% 8.6% 8.6% 7.8% 6.7% 8.7% 8.6% 8.5% 7.7% 6.4% Some college 4.4% 4.1% 4.4% 4.0% 3.9% 4.4% 4.0% 4.4% 4.0% 3.8% At least college 2.8% 2.6% 2.6% 2.6% 2.0% 2.8% 2.6% 2.6% 2.5% 2.0% White 4.7% 4.3% 4.1% 3.5% 2.7% 4.7% 4.3% 4.1% 3.4% 2.6% African-American 13.1% 11.9% 10.9% 9.0% 7.3% 13.0% 11.8% 10.7% 8.8% 6.9% Hispanic 20.8% 17.4% 16.7% 14.5% 12.0% 20.7% 17.4% 16.6% 14.3% 11.8% Other 11.8% 9.8% 8.3% 7.8% 5.9% 11.8% 9.7% 8.1% 7.7% 5.8% Married 4.7% 4.5% 4.7% 4.1% 3.4% 4.6% 4.5% 4.6% 4.0% 3.3% Widowed 7.7% 7.1% 6.8% 6.4% 5.5% 7.7% 7.1% 6.8% 6.3% 5.3% Divorced 14.5% 12.6% 11.1% 9.8% 7.7% 14.5% 12.6% 11.0% 9.5% 7.3% Never married 18.3% 16.1% 15.3% 13.3% 11.4% 18.3% 16.1% 15.3% 13.2% 11.1% Notes: All numbers are in percent. Estimates based on Urban Institute s DYNASIM3 model. See text for innovation specifics. That is, the benefit increases particularly serve the intended target audiences. And the declines in poverty rates in some instances occur earlier for the target audiences than for their counterparts (table 1). The poverty rates for African Americans and Hispanics are somewhat lower than they otherwise would be starting in 2025, while the decline for Whites does not show up until 2055 (table 1). The targeted impact of our proposed changes is also noticeable in the absolute and relative benefit changes (table 2). Women, people never married, communities of color, those with less than high school degrees, and low-income families see larger absolute increases in their Social Security benefits if our proposed changes were enacted, than men, Whites, and those with more education would see. By 2065, benefits for women, for instance, would be $140 (in 2015 dollars) higher than without our proposed changes. And the respective change for divorced people is $153 (table 2). The relevant increase for African Americans is $138, for those without high school degrees $176, and $208 for the lowest fifth of earners (table 2). In comparison, men would see a benefit increase equal to MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 11

TABLE 2. Dollar and Percent Change in Average Per Capita Annuity Income of People Age 62 and Older by Selected Characteristics and Selected Years 2015 Dollars Percent Year 2025 2035 2045 2055 2065 2025 2035 2045 2055 2065 All $5.0 $14.5 $44.6 $89.3 $124.7 0.0% 0.0% 0.1% 0.1% 0.2% Women $4.2 $12.3 $44.9 $95.7 $139.7 0.0% 0.0% 0.1% 0.1% 0.2% Men $6.0 $17.0 $44.3 $82.0 $107.8 0.0% 0.0% 0.1% 0.1% 0.1% Less than h.s. $6.0 $18.3 $58.2 $122.4 $176.0 0.0% 0.1% 0.2% 0.4% 0.5% High school $5.7 $15.5 $48.3 $99.9 $139.2 0.0% 0.0% 0.1% 0.2% 0.2% Some college $4.9 $13.6 $39.7 $85.5 $115.9 0.0% 0.0% 0.1% 0.1% 0.2% At least college $3.8 $12.6 $39.2 $71.8 $103.2 0.0% 0.0% 0.0% 0.1% 0.1% White $4.9 $13.9 $41.8 $84.3 $119.8 0.0% 0.0% 0.0% 0.1% 0.1% African-American $4.4 $13.9 $52.4 $113.1 $138.2 0.0% 0.0% 0.1% 0.2% 0.2% Hispanic $6.5 $17.3 $44.3 $81.6 $121.3 0.0% 0.0% 0.1% 0.1% 0.2% Other $4.0 $15.6 $54.8 $103.3 $141.3 0.0% 0.0% 0.1% 0.1% 0.2% Married $7.1 $18.9 $54.8 $100.8 $131.7 0.0% 0.0% 0.1% 0.1% 0.2% Widowed $1.9 $8.4 $28.9 $68.8 $117.2 0.0% 0.0% 0.0% 0.1% 0.1% Divorced $2.4 $12.7 $48.1 $103.0 $153.7 0.0% 0.0% 0.1% 0.1% 0.2% Never married $2.2 $6.3 $25.9 $64.7 $91.7 0.0% 0.0% 0.0% 0.1% 0.1% Bottom quintile of lifetime earnings distribution Second quintile of lifetime earnings distribution Third quintile of lifetime earnings distribution Fourth quintile of lifetime earnings distribution Top quintile of lifetime earnings distribution $6.9 $18.5 $70.5 $151.3 $207.5 0.0% 0.1% 0.4% 0.7% 0.9% $7.1 $20.0 $52.9 $101.4 $140.4 0.0% 0.1% 0.1% 0.3% 0.3% $4.5 $14.0 $41.7 $79.8 $112.1 0.0% 0.0% 0.1% 0.1% 0.2% $4.0 $11.6 $32.5 $63.4 $92.0 0.0% 0.0% 0.0% 0.1% 0.1% $2.5 $8.5 $25.4 $50.3 $71.6 0.0% 0.0% 0.0% 0.0% 0.0% Notes: All dollar estimates in 2015 dollars. Estimates based on Urban Institute s DYNASIM3 model. See text for innovation specifics. 12 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

$108, married people $132, Whites $120, those with college degrees $103, and those in the highest fifth of lifetime earners would see an increase of $75 by 2065 (table 2). The difference in relative changes is even more pronounced since the target audiences have lower benefits than their counterparts to begin with. The extra dollars amount to a permanent benefit increase equal to 0.2 percent for women by 2065, 0.2 percent for divorced people, 0.2 percent for communities of color, 0.5 percent for those without high school degrees, and 0.9 percent for those in the lowest fifth of the earnings distribution (table 2). In comparison, the extra benefits amount to 0.1 percent of benefits for men, married people, Whites, and those with college degrees (table 2). They also constitute no measurable increase on average in benefits for people in the top earnings quintile (table 2). The results so far suggest that our proposed changes would make Social Security more progressive. Breaking down the relative benefit gains by benefit levels shows this point clearly (table 3). Our innovations result in the largest benefit gains for women, communities of color, and people with less than high school degrees at the 10th percentile of the benefit distribution. Women at that benefit level would experience a 2 percent increase, divorced people an increase equal to 4.7 percent, African Americans a 2.3 percent gain, and those without a high school degree a 3.1 percent uptick (table 3). These increases are higher than those for men, married people, Whites, and college graduates at this income level. They are also higher than the relative gains expected at other income levels (table 3). That is, our innovations not only are well targeted, but they also improve a key feature of Social Security its progressiveness. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 13

TABLE 3. Percent Change in Per Capita Annuity Income among People Age 62 and Older by Selected Characteristics for 2065 Year Mean 5th Percentile 10th Percentile 25th Percentile Median 75th Percentile 90th Percentile 95th Percentile 99th Percentile All 0.2% 2.3% 1.8% 0.5% 0.2% 0.1% 0.0% 0.0% 0.0% Women 0.2% 1.6% 2.0% 0.7% 0.2% 0.1% 0.0% 0.0% 0.0% Men 0.1% 1.4% 1.4% 0.5% 0.3% 0.0% 0.0% 0.1% 0.0% Less than h.s. 0.5% 0.0% 3.1% 4.1% 0.7% 0.6% 0.1% 0.3% 0.1% High school 0.2% 0.5% 2.6% 0.7% 0.4% 0.2% 0.0% 0.0% 0.0% Some college 0.2% 1.5% 0.6% 0.5% 0.2% 0.1% 0.1% 0.0% 0.0% At least college 0.1% 1.1% 0.4% 0.2% 0.1% 0.0% 0.1% 0.0% 0.0% White 0.1% 1.5% 0.8% 0.4% 0.2% 0.1% 0.0% 0.0% 0.0% African-American 0.2% 0.3% 2.3% 0.5% 0.2% 0.4% 0.0% 0.1% 0.0% Hispanic 0.2% 0.0% 1.9% 0.8% 0.3% 0.1% 0.1% 0.2% 0.1% Other 0.2% 4.9% 4.1% 0.8% 0.3% 0.0% 0.0% 0.0% 0.0% Married 0.2% 3.2% 1.9% 0.7% 0.3% 0.1% 0.1% 0.1% 0.0% Widowed 0.2% 2.8% 1.6% 0.5% 0.3% 0.1% 0.0% 0.0% 0.0% Divorced 0.3% 7.8% 4.7% 1.0% 0.3% 0.1% 0.1% 0.1% 0.0% Never married 0.2% 1.7% 4.6% 1.2% 0.4% 0.2% 0.2% 0.0% 0.0% Bottom quintile of lifetime earnings distribution Second quintile of lifetime earnings distribution Third quintile of lifetime earnings distribution Fourth quintile of lifetime earnings distribution Top quintile of lifetime earnings distribution 0.9% 0.0% 5.1% 3.6% 1.3% 0.6% 0.6% 0.2% 0.2% 0.3% 1.3% 0.5% 0.6% 0.5% 0.3% 0.1% 0.2% 0.0% 0.2% 0.3% 0.4% 0.3% 0.2% 0.2% 0.1% 0.0% 0.0% 0.1% 0.3% 0.1% 0.2% 0.1% 0.1% 0.1% 0.1% 0.0% 0.0% 0.1% 0.2% 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% Notes: Estimates based on Urban Institute s DYNASIM3 model. See text for innovation specifics. 14 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

OPTIONS TO PAY FOR ADDITIONAL BENEFITS Better benefits could also mean potentially rising long-term Social Security deficits. The question then arises, Could we pay for the additional benefits and help to stabilize benefits for everybody by shrinking Social Security s long-term deficits? Broadening the tax base to tax capital income, even at a tax rate below the combined payroll tax rate of 12.4 percent, will likely pay for the proposed innovations and for part of the expected long-term financing shortfall. Applying, for example, a tax rate of 6.2 percent to investment incomes interest and dividend income as well as realized capital gains above $200,000 per year starting in 2017 (SSA, 2016a) would pay for the innovations and reduce the long-term deficit. The net changes to income from both benefit and revenue changes once again highlight the progressive nature of our proposal. The net income changes are positive for lowerincome people, but negative for higherincome ones (table 4). And the declines for people in the second and third earnings quintiles are on average fairly modest. We should also note, as we discuss below, that our proposal extends the solvency of the Social Security trust funds for a few years, thus potentially delaying other adverse changes for all beneficiaries. In fact, on average payable benefits are probably more than 5 percent higher over the long-term due to our proposals, based on the numbers in table 5 below. This increase is more than enough to offset the net declines in net incomes, which approximately amount to a decline of not more than 1.2 percent of promised benefits. 11 The simulation results further show that the additional revenue will cover the new benefits and contribute to a reduction of Social Security s long-term deficit. The difference between the income receipts by Social Security and its expenditures will be smaller in every year under our proposal than under current law. In 2065, for instance, the gap between receipts and expenditures would be 8.5 percent of payroll instead of 11.8 percent (table 5). This constitutes a sizable decline in the annual financial gap for Social Security in the long term. Specifically, Social Security s actuarial deficit over the next 75 years would be 2 percent of earnings instead of 2.8 percent of payroll without the changes to benefits and revenue an improvement of 27 percent 11 Data based on simulations, not shown here. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 15

TABLE 4. Change in Average Net Per Capita Cash Income of People Age 62 and Older by Selected Characteristics and Selected Years (in 2015 Dollars) Compared to Current Law, Promised Benefits Year 2025 2035 2045 2055 2065 All $(170.6) $(293.7) $(396.1) $(592.3) $(610.6) Women $(169.3) $(276.6) $(353.3) $(470.8) $(548.8) Men $(172.1) $(313.3) $(444.9) $(729.2) $(680.5) Less than h.s. $(3.4) $(41.5) $8.3 $15.6 $15.3 High school $(44.2) $(109.1) $(199.9) $(277.5) $(307.5) Some college $(136.9) $(243.4) $(256.2) $(443.8) $(478.3) At least college $(426.2) $(613.3) $(804.1) $(1,127.3) $(1,091.8) White $(215.6) $(378.3) $(527.2) $(699.8) $(792.2) African-American $(25.6) $(90.2) $(151.2) $(235.3) $(297.2) Hispanic $(40.5) $(87.7) $(119.5) $(563.7) $(426.2) Other $(141.1) $(224.8) $(337.5) $(470.5) $(418.5) Married $(195.8) $(297.8) $(403.8) $(518.6) $(576.3) Widowed $(123.3) $(276.3) $(382.2) $(488.8) $(628.4) Divorced $(129.6) $(285.9) $(344.9) $(904.4) $(667.4) Never married $(172.9) $(312.6) $(433.8) $(713.1) $(658.2) Bottom quintile of lifetime earnings distribution Second quintile of lifetime earnings distribution Third quintile of lifetime earnings distribution Fourth quintile of lifetime earnings distribution Top quintile of lifetime earnings distribution $3.8 $(0.3) $54.1 $104.6 $114.8 $(17.6) $(13.8) $(23.4) $(0.5) $(61.4) $(31.9) $(67.3) $(85.7) $(131.4) $(232.7) $(111.1) $(194.2) $(279.3) $(407.6) $(558.3) $(696.3) $(1,193.2) $(1,646.5) $(2,526.6) $(2,315.6) Notes: All dollar estimates in 2015 dollars. Estimates based on Urban Institute s DYNASIM3 model. See text for innovation specifics. Numbers in parentheses denote negative values. 16 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

in Social Security s long-term finances. The combination of benefit and revenue changes extends the life of the Social Security trust funds from 2034 to 2037. Moreover, the share of promised benefits that can be paid for with expected payroll tax revenue increases, even after the Social Security trust funds are exhausted. In 2045, for instance, payroll taxes can pay 84.1 percent of promised benefits with our changes, instead of 79.1 percent without the changes (table 5). By 2065, Social Security could still pay 81.6 percent of promised benefits rather than 76.8 percent (table 5). Put differently, our proposal not only improves benefits in a targeted manner, but it also helps to stabilize expected benefits for all workers. TABLE 5. Indicators of Fiscal Changes for Select Years No heading 2025 2035 2045 2055 2065 Annual difference between receipts and expenditures relative to payroll (current law) Annual difference between receipts and expenditures (with innovations) Share of promised benefits payable with tax revenue (current law) Share of promised benefits payable with tax revenue (with innovations) -1.3% -4.1% -5.9% -8.5% -11.8% -0.6% -2.8% -4.0% -5.8% -8.5% 100.0% 77.5% 79.1% 78.3% 76.8% 100.0% 100.0% 84.1% 83.9% 81.6% Notes: All numbers are in percent. Estimates based on Urban Institute s DYNASIM3 model. See text for innovation specifics. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 17

CONCLUSION Workers have seen substantial increases in the volatility of their earnings. People s incomes have become less stable due to added caregiving responsibilities, longer unemployment spells, and fewer career opportunities for people with less education. We propose to make it easier for people to earn credits toward Social Security benefits in those situations caring for others, being unemployed, and pursuing job training. The benefit improvements are well targeted, as they are larger for women than for men, higher for communities of color than for Whites, and greater for people with less education than for those with more education. We further propose to pay for these benefit improvements and for part of Social Security s long-term financial shortfall by imposing a payroll tax on capital income for higherincome earners. The result of these combined changes is a more progressive and financially more viable social insurance system. It is important for us to reiterate that our proposed changes should be considered an important part of a larger package toward countering increasing income uncertainty for many Americans. People will also need stronger jobs, better contemporaneous benefits such as paid family leave and unemployment insurance benefits, and complementary Social Security benefits improvements to ensure the dignity of an above-subsistence income to all retirees. In essence, policy makers wanting to address the rising income volatility will need to pursue a comprehensive reform approach, of which our proposed changes are an integral part. 18 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

APPENDIX Formula for amount of caregiving credit relative to earnings Earnings Level in a Given Year Credit Amount Description 0.6 * AWI 0.6 * AWI Flat dollar amount equal to 60 percent of AWI for those earning 60 percent of AWI or less > 0.6 * AWI & AWI (Earnings/AWI + 0.05 ((AWI earnings)/awi) * 0.125) * AWI Gradual increase in percentage of AWI from 60 percent to 105 percent of AWI for those earning between 60 percent and 100 percent of AWI > AWI 1.05 * earnings Top-up equal to five percent of earnings for those earning at least AWI Notes: AWI stands for average wage index. Earnings refers to a taxable earnings up to the taxable maximum in a given year. MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 19

REFERENCES Boushey, H. 2016. Finding Time. Cambridge, MA: Harvard University Press. Bureau of Labor Statistics (BLS). 2016a. Labor Force Characteristics by Race and Ethnicity: 2015. Washington, DC: BLS. Bureau of Labor Statistics (BLS). 2016b. Labor Force Statistics from the Current Population Survey. Table A-36. Unemployed Persons by Age, Sex, Race, Hispanic or Latino Ethnicity, Marital Status, and Duration of Unemployment. August 2015. Washington, DC: BLS. Chiteji, N. & Hamilton, D. 2002. Family Connections and the Black-White Wealth Gap among the Middle Class. Review of Black Political Economy, Vol. 30. No. 1: 9 27. Deutsche Rentenversicherung (DRV). 2016. Jeder Monat Zaehlt. Berlin, Germany: DRV. Available at http://www.deutscherentenversicherung.de/cae/servlet/ contentblob/232688/publicationfile/52920/ rente_jeder_monat_zaehlt.pdf. Dynan, K., Elmendorf, D. & Sichel, D. 2007. The Evolution of Household Income Volatility. FEDS Paper No. 2007-61. Washington, DC: Federal Reserve. Ellis, R. & Simmons, T. 2014. Coresident Grandparents and Their Grandchildren: 2012. Census Paper No. 20-576. Washington, DC: US Census Bureau. Favreault, M. 2008. A New Minimum Benefit for Low Lifetime Earners. Washington, DC: National Academy of Social Insurance. Available at https://www.nasi.org/sites/ default/files/research/melissa_favreault_ January_2009_Rockefeller.pdf. Favreault, M. M., Mermin, G. B. T. & E. Steuerle, C. 2006. Minimum Benefits in Social Security. PPI Discussion Paper #2006-17. Washington, DC: AARP Public Policy Institute. Available at http://assets.aarp. org/rgcenter/econ/2006_17_socsec.pdf. Fetterolf, C. & Rudman, L. 2014. Gender Inequality in the Home: The Role of Relative Income, Support for Traditional Gender Roles, and Perceived Entitlement. Gender Issues, Vol. 31, No. 3 4: 219 37 Fryer, R., Pager, D. & Spenkuch, J. 2011. Racial Disparities in Job Finding and Offered Wages. Working Paper. Cambridge, MA: Harvard University. 20 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD

Garber, J. 2016. Is There a Racial Volatility Gap? Financial Security Program, Aspen Institute. August 1. Government Accountability Office (GAO). 2007. Women Face Challenges in Ensuring Financial Security in Retirement. Washington, DC: GAO. Hamilton, D., Austin, A. & Darity, W., Jr. 2011. Occupational Segregation and the Lower Wages of Black Men. Economic Policy Institute Briefing Paper. Washington, DC: Economic Policy Institute. Hamilton, D. 2009. Race, Wealth, and Intergenerational Poverty: There will Never Be a Post-Racial America if the Wealth Gap Persists. The American Prospect, Special Report (August). Available at http://prospect.org/article/race-wealthand-intergenerational-poverty. Hamilton, D. & Darity, W., Jr. 2010. Can Baby Bonds Eliminate the Racial Wealth Gap in Putative Post-Racial America? Review of Black Political Economy, Vol. 37, Nos. 3/4: 207 16. Hardy, B. 2017. Income Instability and the Response of the Safety Net. Contemporary Economic Policy Vol. 35, No: 2 312 330. doi:10.1111/coep.12187 Hardy, B. & J. Ziliak. 2014. Decomposing Rising Income Volatility: The Wild Ride at the Top and Bottom. Economic Inquiry, Vol. 52, No. 1: 459 76. Herd, P. 2006. Crediting Care or Marriage? Reforming Social Security Family Benefits. The Journals of Gerontology: Series B, Vol. 61, No. 1: S24-S34. Herd, P. 2009. Crediting Care in Social Security: A Proposal for an Income Tested Care Supplement. Washington, DC: National Academy of Social Insurance. Available at https://www.nasi.org/usr_doc/pamela_herd_ January_2009_Rockefeller_Project.pdf. Johnson, R. & Lo Sasso, A. 2006. The Impact of Elder Care on Women s Labor Supply. Inquiry, Vol. 43, No. 3: 195 210. Kopczuk, W., Saez, E. & Song, J. 2010. Earnings Inequality and Mobility in the United States: Evidence from Social Security Data Since 1937. Quarterly Journal of Economics, Vol. 125, No. 1: 91 128. Lambert, S., Fugiel, P. & Henly, J. R. August 27, 2014. Precarious Work Schedules among Early-Career Employees in the US: A National Snapshot. Research brief issued by EINet (Employment Instability, Family Well-being, and Social Policy Network) at the University of Chicago. Maume, D. 2008. Gender Differences in Providing Urgent Childcare among Dual-Earner Parents. Social Forces, Vol. 87, No. 1: 273 97. Michaelides, M. & Mueser, P. 2012. Recent Trends in the Characteristics of MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD 21

Unemployment Insurance Recipients. Monthly Labor Review, July: 28 47. Social Security Administration (SSA). 2016a. Estimates of the Financial Effects on Social Security of S. 731, the Social Security Expansion Act, Legislation Introduced on March 12, 2015 by Senator Bernie Sanders. Baltimore, MD: SSA. Social Security Administration (SSA). 2016b. Actuarial Resources. Provisions. Provision D2. Baltimore, MA: SSA. Available at https://www.ssa.gov/oact/solvency/ provisions/charts/chart_run241.html. Weller, C. 2015. Retirement on the Rocks: Why Americans Can t Get Ahead and How New Savings Policies Can Help. New York, NY: Palgrave-MacMillan. Weller, C. & Tolson, M. 2017. Women s Economic Risk Exposure and Savings. Paper presented at the LERA/ASSA annual meetings, Chicago, January. Ziliak, J., Hardy, B. & Bollinger, C. 2011. Earnings Volatility in America: Evidence from Matched CPS. UKCPR Discussion Paper No. 2011-03. Lexington, KY: University of Kentucky, Center for Poverty Research. 22 MORE RETIREMENT STABILITY IN AN UNSTABLE WORLD