Earnings Sharing in Social Security: Projected Impacts of Alternative Proposals Using the MINT Model

Abstract

Changes in American family and work patterns over the past decades have prompted various policy proposals for changing the structure of Social Security benefits. In this article, we use the Social Security Administration's Modeling Income in the Near Term (MINT) microsimulation model to project how Social Security benefit amounts would change in response to incorporating earnings sharing into benefit calculations for the population aged 62 or older in 2030 under three hypothetical policy scenarios. The earnings sharing scenarios modeled in the article would reduce benefits for the majority of individuals, although there are important differences among married, divorced, and widowed individuals. Some groups of men and women would experience increases in Social Security benefits, while some would receive reduced benefits in comparison to current law, particularly widowed individuals. Allowing widows to inherit the earnings records of their deceased husbands would improve their outcomes.

Summary

This article provides policymakers and retirement analysts with insights into the potential distributional effects of incorporating earnings sharing in the calculation of Social Security benefits. Earnings sharing refers to a system whereby the earnings records of married individuals are combined and split equally for each year of marriage for the purpose of calculating each individual's Social Security benefit. Incorporating earnings sharing has been proposed as one way to adapt the Social Security program to socioeconomic and demographic changes over the last few decades, such as changes in women's work and marriage patterns.

Using the Social Security Administration's Modeling Income in the Near Term (MINT) microsimulation model, we estimate the impact of earnings sharing on Social Security benefits for the projected retirement population aged 62 or older in 2030 under three hypothetical policy proposals: earnings sharing with no auxiliary survivor or spousal benefits, earnings sharing with survivor benefits only, and earnings sharing with survivor and spousal benefits. We exclude any additional benefit enhancements from the three proposals to focus on the fundamental effects of basic earnings sharing on future retirees.

In evaluating the desirability of earnings sharing, it is important to consider how the policy would affect different segments of the retiree population. Overall, the earnings sharing scenarios analyzed in this article would lead to benefit decreases relative to current-law Social Security benefits for the majority of future retirees. Results differ, however, among married, divorced, and widowed individuals. For married individuals, over a third would receive Social Security benefit increases, and about half would receive benefit decreases under basic earnings sharing with no auxiliary benefits. The remaining married individuals would experience no change in benefits. Married individuals in one-earner couples would experience more widespread and greater benefit reductions than those in two-earner couples.

For divorced individuals, the majority of men and almost half of women would receive Social Security benefit decreases under earnings sharing with no auxiliary benefits. Only about a tenth of divorced men and two-fifths of divorced women would receive benefit increases. Among those receiving benefit decreases, divorced men would receive an average decrease of about 11 percent and divorced women about 22 percent. For widowed individuals, earnings sharing with no auxiliary benefits would lower the Social Security benefits of the vast majority of men and women. The reduction would average a quarter of the current-law benefits for widows and a fifth of the benefits for widowers.

The introduction of survivor benefits under basic earnings sharing would generate very modest changes, mainly in reducing the extent of benefit decreases for widows and divorced surviving spouses. Providing spousal benefits under an earnings sharing framework would have almost no additional impact. Allowing widows to inherit the earnings records of their deceased husbands would improve their outcomes.

Introduction and Background

Changes in the American family and work patterns over the past decades have prompted various policy proposals to change Social Security's auxiliary benefit system.¹ This article analyzes earnings sharing as an alternative to the current method of calculating Social Security benefits. Using the MINT model, we examine the potential distributional effects of incorporating earnings sharing into benefit calculations for the projected retirement population in 2030 under three hypothetical policy scenarios.

Under the current Social Security system, benefit eligibility is contingent on an individual's own earnings history and on his or her marital history and the earnings records of current and previous spouses. The current system bases retirement benefits on the worker's own earnings and provides auxiliary benefits to dependents and survivors of workers. Auxiliary benefits provide monthly payments to qualified spouses, ex-spouses, and survivors of insured workers. A spouse is entitled to up to half the benefit, and a survivor is entitled to as much as the full benefit, that is due the higher earning spouse (usually the husband).² Although gender-neutral, auxiliary benefits are especially important to women because women tend to have lower lifetime earnings, have fewer years in the workforce, and live longer than men in retirement (Sandell and Iams 1997; Tamborini and Whitman 2007; Weaver 1997, 2002).

Earnings sharing equally divides the combined Social Security earnings of married couples in order to calculate each spouse's benefit. In years when an individual is not married, the individual's own earnings record is used. The earnings sharing approach reflects the assumption that economic resources acquired during a marriage, including earnings, are shared equally regardless of the household division of labor. In the case of multiple marriages or divorce, the sharing would occur with different spouses over the lifetime during each period of marriage. In the most basic form of earnings sharing, a spouse would not be eligible for auxiliary benefits (see, for example, Forman 2006).

Although opponents of earnings sharing point to sharp benefit reductions among some vulnerable groups, and to the cost and difficulty of implementation, proponents often focus on earnings sharing as a means to address inequities in the current Social Security system. A prominent equity-related concern is the treatment of one-earner married couples relative to single persons and two-earner married couples (Favreault, Sammartino, and Steuerle 2002b; Steuerle and Bakija 1994). For example, a one-earner couple with the same total lifetime earnings as a two-earner couple receives higher total benefits³ under the current auxiliary benefit system.⁴

To illustrate this point, Table 1 displays three stylized retired couples. Each couple has the same average lifetime monthly earnings of $1,000, but different intrahousehold earnings profiles. In Couple A, the wife did not work and the husband worked. In Couple B, both the wife and the husband worked, and the wife earned one quarter of her husband's average monthly earnings. In Couple C, the wife and the husband both worked, each with the same average monthly earnings. Although each of the three couples has the same average monthly combined earnings of $1000, one-earner Couple A receives a higher total-couple Social Security benefit ($1,128) than two-earner Couple B ($1,032) and two-earner Couple C ($900).⁵ Couple A receives a higher total-couple benefit because the nonworking wife receives an auxiliary spousal benefit of half the working husband's retired-worker benefit of $752.

Earnings sharing has been suggested as a way to equalize benefits between one-earner and two-earner couples. In Table 1, if we assume that each couple was married for 35 years, both spouses are the same age, all earnings occurred during marriage, and all earnings were shared, then the benefits received by two-earner Couple C would equal the benefits received by one-earner Couple A and two-earner Couple B.⁶ Under earnings sharing, couples with the same total lifetime earnings generally would receive the same benefits regardless of their individual earnings profiles, all things being equal.⁷

Earnings sharing proposals have also been driven by concerns about benefit adequacy, particularly for growing subpopulations such as divorced women who were married for fewer than 10 years.⁸ Under earnings sharing, divorced women whose marriages were too short to qualify for divorced spouse or survivor benefits could see benefit increases if their ex-husbands' earnings were higher than their own during the period of marriage.⁹

Earnings sharing proposals received considerable attention from policymakers in the 1980s. At that time, a number of studies evaluated the effects of earnings sharing and the transition costs of moving to such a system. The Social Security Administration (SSA), then part of the Department of Health and Human Services (HHS), conducted a broad implementation study on a set of complex earnings sharing proposals debated during the early to mid-1980s.¹⁰ That study modeled a generic version of earnings sharing in which each spouse was credited with half of a couple's combined covered (Social Security) earnings for each year of marriage. The generic model, however, included "offset reductions in benefits for survivors compared to present law;" that is, surviving spouses and surviving divorced spouses could "inherit the total amount of the deceased spouse's covered annual earnings for each year of marriage and. . .add this amount to his or her own earnings" (HHS 1985a, XIV). Thus survivors would be credited for combined earnings for each year they had been married, but spouses would only be credited for half of the combined earnings for each year of marriage to a still-living spouse. Overall, the study found mixed results. Benefit declines, for example, were documented among almost half of couples (especially one-earner couples), over two-fifths of widows, and over half of divorced men, while benefit increases were found among some individuals from a two-earner couple and around half of divorced women.¹¹

Although receiving less attention in more recent years, earnings sharing continues to be cast as a policy alternative. Several recent studies, such as Favreault and Steuerle (2007) and Schwabish, Simpson, and Topoleski (2007), examine earnings sharing as part of a broader set of policy packages to address the changing retirement needs of American families. In these studies, as in others, earnings sharing is viewed as a way to address inequities between one-earner and two-earner married couples under the current Social Security program. Moreover, like the earnings sharing proposals examined in the 1980s, the proposals often include auxiliary benefits and other enhancements, such as an increased minimum benefit or caregiving credits to offset benefit decreases overall and to protect groups such as widows who may otherwise experience benefit reductions under earnings sharing.¹² A consequence of including enhancements, however, is that it can make policy proposals very complex, and ascertaining the benefit increases and decreases attributable to earnings sharing is difficult (Fierst 1990).

This article reassesses how the Social Security benefits of future retirees would change in response to earnings sharing without enhancements in Social Security's benefit structure. Enhancements would make an earnings sharing plan more politically viable, but removing them for analytical purposes makes it easier for policymakers and advocates to ascertain the fundamental distributional effects of earnings sharing. Three hypothetical earnings sharing options are examined, with distributional effects projected for the population aged 62 or older in 2030. The article does not advocate or oppose the policy options examined herein. A brief description of the data and methodology follows. The subsequent section reports the results. The concluding section contains a summary of the findings and suggestions for possible future work.

Data and Methodology

To estimate the potential distributional effects of earnings sharing on future retiree populations, we use SSA's MINT model.¹³ Developed by SSA's Office of Research, Evaluation, and Statistics with assistance from the Brookings Institution, the RAND Corporation, and the Urban Institute, MINT is a microsimulation model that uses observed and estimated population characteristics to project the demographic characteristics and economic status of future retirees. MINT is a powerful tool for evaluating future aged populations and permits distributional analyses of different policy changes across heterogeneous populations, accounting for socioeconomic and demographic changes among more recent cohorts.¹⁴ The model is based on nationally representative microdata from the 1990–1993 and 1996 panels of the Census Bureau's Survey of Income and Program Participation (SIPP) matched to SSA administrative records.¹⁵

Using the matched data, MINT follows a series of sophisticated techniques that involve systematic modeling of income determinants to project future retirees' Social Security benefits and other retirement income, changes in workforce participation, longevity, and other factors such as date of retirement, marital status changes, and education patterns (Toder and others 2002, II-10). For a thorough description of MINT's methodology, readers should consult Butrica, Iams, Moore, and Waid (2001); Smith, Cashin, and Favreault (2005); and Toder and others (2002).

To date, MINT has not been used to evaluate the distributional consequences of earnings sharing. To calculate Social Security spousal and survivor benefit amounts, MINT identifies characteristics of current, former, and future spouses, and calculates shared earnings based on their lifetime projected earnings. For the purposes of this article, earnings sharing is defined as the combined Social Security earnings of a couple, which is evenly split between the spouses in each year of marriage to calculate Social Security benefits. A person retains his or her own earnings in each year he or she is unmarried.¹⁶

We simulate three policy proposals, all of which use earnings sharing to calculate Social Security benefits in place of the current-law benefit calculation. The first policy proposal (P1) eliminates all auxiliary benefits. This option adheres to the most basic form of earnings sharing, which "would eliminate the current system of benefits for workers and spouses (or surviving spouses) and instead credit each spouse with half of a couple's total covered earnings for each year of marriage" (HHS 1985a, XIII). The second policy proposal (P2) retains survivor benefits, but calculates such benefits under earnings sharing rather than current law. Under P2, the survivor benefit would raise the benefit of the survivor to the level of the earnings sharing benefit of the deceased spouse if the survivor's own earnings sharing benefit is lower than the deceased spouse's earnings sharing benefit. The third policy proposal (P3) keeps the current structure of spousal and survivor benefits intact, but again calculates those benefits under earnings sharing rather than using the individual earnings of the highest earner. Under P3, the spousal supplement would raise the earnings sharing benefit of the lower-earning spouse to the level of half the earnings sharing benefit of the higher-earning spouse.

Retaining spousal and survivor benefits adds a layer of protection for groups who may receive reduced benefits under earnings sharing. However, unlike much of the previous literature, this study bases auxiliary benefits on shared earnings, which provides a clear estimate of the distributional effect of an earnings sharing system, both with and without auxiliary benefits, and is more consistent with the concept of earnings sharing than other types of enhancements. Since spousal and survivor benefits are calculated using earnings shared during years of marriage, auxiliary benefits could be lower than under the current system.

The simulations measure the impact of the three alternatives on average Social Security benefits for retirees aged 62 or older in 2030 (born 1926–1968). Age 62 is chosen since that is the age at which individuals are entitled to receive retired-worker benefits and also at which spouses of retired workers are eligible for benefits. Using the 62-or-older group in 2030 allows the analysis of the effects on a retiree population mainly consisting of members of the baby-boom generation, a cohort at the forefront of sharp demographic changes in the American population, such as women's increased labor force participation and downward marriage trends.¹⁷ We focus on the impact of earnings sharing on benefits in 2030 rather than on lifetime benefits. A person's benefit can increase or decrease over time depending on changes in marital status and earnings, so the effect of a policy change on benefits in a single year can differ from the effect on lifetime benefits.

We assess the likelihood of receiving benefit increases and decreases for the entire population, married individuals, divorced men and women, and widowers and widows. We also distinguish between individuals in one-earner and two-earner married couples.¹⁸ Individuals are identified as married, divorced, or widowed according to their current marital status in 2030.¹⁹ Individuals in 2030 who were never married are excluded from the analysis since never-married individuals are not affected by earnings sharing. Individuals projected to be eligible for disabled-worker benefits are also excluded from the analysis because of the incomplete nature of their earnings histories. Our analysis assumes that all beneficiaries would have had the opportunity to share earnings over their entire working and married lifetimes. The results are weighted to be nationally representative.

Several limitations are worth noting. Because the MINT population is based on the SIPP survey panels, MINT projections contain sampling errors. More important is the uncertainty related to projection error, which reflects differences between MINT estimates and future trends. For these reasons, small differences in our results should be viewed with caution. Another noteworthy point is that this MINT analysis does not assume any type of behavioral response to policy changes. Finally, it is outside the article's scope to consider the effects of earnings sharing proposals on system financing and the transition or administrative costs involved in moving from the existing system to an earnings sharing system. At a time when the system faces long-term financial challenges, the implications of policy alternatives on Social Security's finances are an important consideration.

Results

Overall Population 62 or Older

Table 2 presents results for the overall population aged 62 or older in 2030. Note that "no change in benefits" is defined as having projected benefits change less than 1 percent from current law. Benefits must change by 1 percent or more from current law to be defined as increases or decreases. Benefit amounts are expressed in 2005 dollars. For divorced and widowed individuals, benefits are annual individual benefits. For married individuals, benefits are half the combined annual couple benefit. Only those eligible for benefits under current law are included in the tables. For married beneficiaries, the spouse must also be eligible for benefits under current law to be included in the tables.²⁰

Overall, the three earnings sharing policies would reduce Social Security benefits for the majority of individual beneficiaries. P1 would reduce benefits by 8 percent, and P2 and P3 would each reduce benefits by 4 percent. Approximately 60 percent of individuals would receive benefit reductions, almost 30 percent would receive benefit increases, and 11 percent to 14 percent would experience no change in benefits. For individuals who receive benefit increases, benefits would increase, on average, by 8 percent. For individuals who experience benefit reductions, benefits would decrease, on average, by 11 percent to 17 percent.

Married Individuals

Table 3 presents the distributional effects of the earnings sharing proposals on couple benefits for individuals who are married in 2030. Since earnings sharing essentially redistributes earnings and benefits within couples, the effects of the proposals for married individuals are gauged from a couple perspective. The combined couple benefits are halved in Tables 3 and 4 to make the results for married individuals more easily comparable to the results for divorced and widowed individuals. We refer to this split couple benefit as the per capita benefit. Both the husband and wife must be eligible for benefits under current law to be included in the table.

Fourteen percent of married individuals would be unaffected by P1, 37 percent would receive increased benefits, and 49 percent would receive reduced benefits. The average annual current-law per-capita benefit in 2030 would be slightly higher for married individuals with increases ($15,221) than those with decreases ($13,589). The average increase would be 7 percent and the average reduction would be 8 percent of benefits. Thus, P1 slightly increases benefits for over a third of married individuals and slightly decreases benefits for about half of married individuals.

The addition of survivor benefits would have little effect on married individuals, so results for P2 are not shown. Auxiliary spousal benefits in earnings sharing proposal P3 would only slightly alter the impact on married individuals in terms of the percentage with increases and decreases, as well as the size of the average increase and decrease.²¹

One-earner versus two-earner married couples. As previously discussed, a major rationale for earnings sharing is to improve the equity of benefits between one-earner and two-earner couples. Under current law, a one-earner couple receives higher benefits than a two-earner couple with the same lifetime earnings. In the context of this study, we therefore might expect that one-earner couples would experience greater reductions in benefits than two-earner couples.

Table 4 illustrates the effects of the three earnings sharing proposals for married individuals in one-earner and two-earner couples, shedding light on the complex interactions between marriage patterns, earnings histories of spouses, and Social Security program rules.²² Overall, the prevalence of benefit increases would be substantially lower among individuals in one-earner married couples. P1 would increase benefits for about a quarter of individuals in one-earner couples and for about two-fifths of individuals in two-earner couples.²³ Increases for one-earner couples could result when both spouses have low earnings and one spouse does not have enough earnings to qualify for retired-worker benefits. Under current law the latter spouse receives spousal benefits, but under P1, both spouses would qualify for retired-worker benefits, so the total benefits received by the couple could increase. The percentage with increases would be similar for individuals in both types of two-earner married couples (those in which both are eligible for retired-worker benefits and those in which one spouse is dually entitled and receives higher spousal benefits under current law).

As expected, benefit reductions would be more widespread and greater for individuals in one-earner married couples than for those in two-earner married couples. About two-thirds of individuals in one-earner married couples would experience a reduction in benefits under P1, and their average benefit decrease would be 20 percent. In contrast, about two-fifths of individuals in two-earner married couples with only retired-worker benefits and about half of those with dual entitlement would experience decreases in their average benefits, with relatively small decreases of 5 percent and 7 percent, respectively.

The results for P2 would be the same as for P1 because the analysis is restricted to married individuals in their first marriage, and therefore survivor benefits do not apply to this subsample. Interestingly, proposal P3, which includes spousal benefits, would have little effect on couple benefits beyond that of P1. Adding spousal benefits has a fairly small effect because the spouse's benefits would be based on the higher earner's shared earnings (when married) and his or her own earnings (when not married), rather than solely on the higher earner's own earnings history, as calculated under current law.

Examining the results for married individuals by earner status suggests important differences between one-earner and two-earner married couples. Individuals in two-earner married couples would be more likely to experience benefit increases, and less likely to experience benefit decreases, than those in one-earner married couples. Under P3, the benefit reductions for individuals in one-earner married couples would still be much greater than those for individuals in two-earner married couples, even for the dually entitled.

Divorced Individuals

The proposals produce important differences between divorced women and men (Table 5). P1 would increase benefits for a higher share of divorced women than men. Specifically, about 38 percent of divorced women would receive benefit increases averaging 13 percent, while 12 percent of divorced men would receive benefit increases averaging 9 percent.

However, P1 would produce more benefit reductions than increases for both women and men. P1 would reduce benefits for a greater share of divorced men, but women would face larger reductions: 77 percent of divorced men would experience reductions averaging 11 percent, compared with 45 percent of divorced women experiencing a reduction averaging 22 percent. These changes reflect the complex benefit calculation under the earnings sharing proposals, in which the final benefit reflects the divorced individual's own earnings when unmarried and shared earnings during marriage. Some divorced women could receive lower benefits under the earnings sharing proposals because they would no longer benefit from the postmarriage earnings of their often higher-earning ex-husband, as under current law. Divorced men would tend to receive benefit decreases more often than women, in part because, even though men's earnings are typically higher than the earnings of their wives, these earnings would be divided under earnings sharing during each year of marriage.

Providing auxiliary benefits under proposals P2 and P3 would only slightly change the distribution of increases and decreases among divorced women and men. However, providing survivor benefits would markedly reduce the relative decrease in benefits for divorced women, from 22 percent under P1 to 9 percent under P2, and would slightly reduce the decrease in benefits for currently divorced men, from 11 percent under P1 to 9 percent under P2. Thus, although similar proportions of divorced women would face benefit decreases under P2 and P1, the average decrease under P2 would be mitigated by the addition of survivor benefits. This is largely because survivor benefits received by the majority of eligible divorced women are based on the shared earnings records of their deceased ex-husbands, which tend to be higher than their own lifetime shared earnings records. Thus, the survivor benefit under P2 yields a higher benefit for many divorced women than the retired-worker benefit under P1; however, such a benefit is still lower than the current-law survivor benefit because the latter is calculated based on the deceased ex-husband's own lifetime earnings history without sharing in years of marriage, which is generally higher than his shared earnings history.

In sum, the majority of currently divorced men and almost half of currently divorced women would receive reduced benefits under the examined earnings sharing proposals. Only about a tenth of currently divorced men and two-fifths of currently divorced women would receive increased benefits.²⁴

Widowed Individuals

Table 6 shows that the effect of the three earnings sharing proposals is very different for widowed individuals than for other subgroups of the elderly population.²⁵ Under P1, benefits would decrease for 93 percent of widows and 95 percent of widowers. Among those who receive reduced benefits, average benefits would decrease by 27 percent for widows and by 19 percent for widowers. Only 5 percent of widows and 3 percent of widowers would receive small increases.

The dramatic reduction in benefits for widowed individuals under P1 is attributed to several factors. Under P1, a widow receives the same benefit before and after the death of her husband. In contrast, under current law, the death of a husband initiates a new survivor benefit based on up to 100 percent of the deceased husband's benefit. Assuming the husband was the higher earner in the couple, earnings sharing would reduce the earnings credited to the husband and increase the earnings credited to the wife. However, the credited earnings of each of the spouses under earnings sharing would be lower than the earnings credited to the husband under current law, thus the wife's benefit under earnings sharing would be lower than the current-law survivor benefit based on her deceased husband's benefit. Similarly, the benefit received by the widower under earnings sharing would be lower than the current-law benefit based on his own nonshared earnings.

The introduction of survivor benefits in proposal P2 would result in a slightly higher percentage of widowed individuals who would experience benefit increases under earnings sharing, and a corresponding slightly lower percentage who would receive benefit decreases. P2 would reduce the average decrease in benefits for those experiencing reductions by 11 percentage points for widows, but only by 4 percentage points for widowers. However, the vast majority of widows would still receive lower survivor benefits under P2 than the current system. This is because the shared earnings record of the deceased husband is often lower than his nonshared earnings record.

The results for P3 are not reported because they would be the same as the results for P2 as, in general, widows and widowers do not receive spousal benefits. Taken together, among the projected retirement-age population in 2030, widowed individuals would mainly experience benefit reductions under these earnings sharing proposals, even under P2, and this decrease would average between about one-sixth and one-quarter of benefits.

The dramatic estimated benefit decreases among widowed individuals has prompted the introduction of earnings sharing proposals with provisions aimed toward offsetting the benefit reductions for survivors, particularly for widows. Even though there is no clear or consistent approach to calculating benefits for survivors under earnings sharing, one plausible approach is to allow survivors to inherit the nonshared Social Security earnings record of their deceased spouses for each year of marriage. Table 7 compares the results for P1 with and without inheritance of earnings for widowed individuals.²⁶ The inheritance proposal reduces the percentage of widows receiving benefit decreases and raises the percentage of widows receiving benefit increases as compared to P1 without inheritance. However, even with the inheritance of earnings, about a third of widows would receive benefit decreases relative to current law.²⁷ This would occur if the deceased husband had higher earnings than the wife before their marriage. Such earnings contribute to the current-law survivor benefit but are omitted from the inherited survivor benefit under earnings sharing.

Conclusions

This article examines the impact of three earnings sharing scenarios on the retirement-age population in 2030 using a recent version of MINT, a microsimulation model that has not previously evaluated the distributional consequences of earnings sharing. The policy alternatives modeled in this article represent basic earnings sharing scenarios. This approach yields more straightforward results than much of the previous literature on earnings sharing, which added a Social Security benefit adjustment or enhancement such as an increased minimum benefit or caregiving credits, in large part to adjust for the sharp decreases that would otherwise be experienced by some groups, especially widows. Although it is understandable why previous analyses included these enhancements, doing so can make it difficult for policymakers and advocates to ascertain the distinct fundamental distributional effects of earnings sharing by changing the distribution of benefit increases and decreases. A more politically viable earnings sharing plan likely would base survivor benefits on the full earnings record of the deceased husband, rather than on shared earnings during periods of marriage.

Overall, these three earnings sharing proposals would lead to a reduction of current-law Social Security benefits for the majority of retirees in 2030. However, important differences exist between married, divorced, and widowed individuals. Nearly half of married individuals would receive lower benefits in 2030. Benefit reductions would be more widespread for married individuals in one-earner couples, and conversely, benefit increases would be more prevalent for those in two-earner couples. Among divorced and widowed individuals there are important differences, with some experiencing benefit increases, but many experiencing benefit decreases. The vast majority of widows and widowers would receive benefit reductions, with widows experiencing greater relative declines than widowers. Although the effect would be more mixed for divorced beneficiaries, the earnings sharing scenarios examined in this article would reduce the benefits of three-fourths of men and nearly one-half of women. Earnings sharing with survivor benefits based on shared earnings would moderately reduce the benefit decreases for widows and divorced surviving spouses. Adding spousal benefits would not substantially alter the distributional effects.

The analysis shows that the three earnings sharing scenarios improve benefit adequacy for some, while reducing it for many others. The results are consistent with prior research which showed that earnings sharing would not improve benefit adequacy for some of the most economically vulnerable groups; instead, many survivors of retired workers and divorced women would be financially worse off under an earnings sharing approach than under current law.

Future work could delve deeper into the extent to which earnings sharing may differentially affect retirees with different socioeconomic and demographic characteristics. There is the need, for example, to further examine the effect of earnings sharing on the progressivity of the Social Security system, namely by focusing on potential changes in benefits across educational and income subpopulations within different marital groups. Analyzing beneficiaries' income level and type of Social Security benefit received could help explain how complex interactions between marital and earnings histories cause certain groups to receive benefit increases and others to receive benefit decreases under earnings sharing. Additional research could also look at the effects of earnings sharing on poverty and lifetime benefits, or further explore the distributional impact of earnings sharing with inheritance. It would also be informative to focus attention on how nonworking spouses or secondary earners fare under earnings sharing, and to examine the cost effects of earnings sharing by separating them from the redistributive effects.

This article does not offer any policy recommendation, and it neither supports nor opposes earnings sharing. Rather, its purpose is to highlight some of the potential distributional effects to take into account when considering a range of Social Security policy alternatives.