Racial and Ethnic Differences in the Retirement Prospects of Divorced Women in the Baby Boom and Generation X Cohorts

by
Social Security Bulletin, Vol. 72, No. 1, 2012

Blacks, Hispanics, and divorced women have historically experienced double-digit poverty rates in retirement, and divorce and other demographic trends will increase their representation in future retiree populations. For these reasons, we might expect an increase in the proportion of economically vulnerable divorced women in the future. This article uses the Social Security Administration's Modeling Income in the Near Term (version 6) to describe the likely characteristics, work experience, Social Security benefit status, and economic well-being of future divorced women at age 70 by race and ethnicity. Factors associated with higher retirement incomes include having a college degree; having a strong history of labor force attachment; receiving Social Security benefits; and having pensions, retirement accounts, or assets, regardless of race and ethnicity. However, because divorced black and Hispanic women are less likely than divorced white women to have these attributes, income sources, or assets, their projected average retirement incomes are lower than those of divorced white women.


Barbara Butrica and Karen Smith are senior research associates at the Urban Institute.

This research was funded by the Social Security Administration (contract no. SS00-06-60113 and order no. SS00-10-31234).

The findings and conclusions presented in the Bulletin are those of the authors and do not necessarily represent the views of the Social Security Administration or the Urban Institute, its trustees, or funders.

Introduction

Selected Abbreviations
MINT6 Modeling Income in the Near Term, version 6
PIA primary insurance amount
SIPP Survey of Income and Program Participation
SSA Social Security Administration
SSI Supplemental Security Income

Since the 1960s, marriage and divorce patterns have changed in dramatic and well-documented ways. Marriage and remarriage rates have plummeted, divorce rates have soared, and marriage durations have shortened (Ahlburg and De Vita 1992; DaVanzo and Rahman 1993; Goldstein 1999; Kreider and Ellis 2011; NCHS 1991; Norton and Miller 1992; Stevenson and Wolfers 2007; Tejada-Vera and Sutton 2010). These trends have more than tripled the share of women aged 65 or older who are divorced, from 3 percent to 11 percent between 1980 and 2009 (Census Bureau 1995, Table 48; 2011, Table 34). Some researchers project that in the future as many as one in five women will be divorced at retirement age (Butrica and Iams 2000; Butrica and Smith 2012).

The trends in marriage and divorce have not been experienced equally across racial groups. Sweeney and Phillips (2004) find that divorce rates stabilized for white women after the mid-1970s, but they have increased somewhat since the late 1980s for black women. In addition, researchers find that marriage will remain nearly universal for whites and Hispanics but much less so for blacks (Goldstein and Kenney 2001; Harrington Meyer, Wolf, and Himes 2005, 2006; Kreider and Ellis 2011; Norton and Miller 1992; Stevenson and Wolfers 2007). Whites are increasingly more likely than blacks to ever marry. Blacks who do marry are more likely than whites and Hispanics to divorce after the first marriage and are less likely to remarry. Among blacks, 49 percent of first marriages ultimately end in divorce, compared with 41 percent for whites and 34 percent for Hispanics (Kreider and Ellis 2011). Among married individuals in 2009, 71 percent of whites had reached their tenth anniversary, while only 59 percent of blacks had.

These trends will undoubtedly affect the composition and economic well-being of future retiree populations. Historically, older divorced women, blacks, and Hispanics have had significantly lower incomes and higher poverty rates than their counterparts. Among today's women aged 65 or older, 20 percent of those who are divorced are poor, compared with 5 percent of those who are married, 18 percent of those who never married, and 15 percent of widows. Poverty rates of older women also vary dramatically by race and ethnicity: 24 percent of blacks and 22 percent of Hispanics are poor, compared with only 11 percent of whites (SSA 2010, Table 11.1).

One-third of Social Security beneficiaries aged 65 or older depend on their benefits for 90 percent or more of their total income and nearly two-thirds of beneficiaries rely on their benefits for 50 percent or more of their total income (SSA 2010, Table 9.A1). However, Social Security eligibility and benefit amounts depend not only on earnings, but also on marital status. A number of studies have recognized the negative impact that recent marriage and divorce trends could have on Social Security benefits for future generations of retired women (Butrica and Iams 2000; Butrica and Smith 2012; Tamborini and Whitman 2007). Tamborini, Iams, and Whitman (2009) compare marital histories in the 1990 and 2004 Surveys of Income and Program Participation (SIPP) and find a modest decline in Social Security spouse and widow benefit eligibility, particularly among black baby boomer women. Likewise, Harrington Meyer, Wolf, and Himes (2005, 2006) analyze the June 1985, 1990, and 1995 Current Population Surveys and find that the proportion of women who will reach age 62 without a 10-year marriage (which is required to qualify for Social Security benefits based on an ex-husband's earnings history) is increasing modestly for whites and Hispanics, but dramatically for blacks.

This article examines the characteristics, work experiences, retirement resources, and economic well-being of future divorced aged women by race and ethnicity in light of the historical trends in marriage and divorce. It updates Butrica and Iams (2003) using a version of the same microsimulation model that incorporates more recent data and updated projection methods.

Among white, black, and Hispanic divorced women at age 70, we find that whites are projected to be the most economically advantaged in retirement, while blacks and Hispanics are projected to be the least economically advantaged. Regardless of race and ethnicity, factors associated with higher retirement incomes include having a college degree; having a history of strong labor force attachment; receiving Social Security benefits; and having pensions, retirement accounts, or assets. However, because black and Hispanic women are less likely than white women to have these attributes, income sources, or assets, their projected average retirement incomes are lower than those of divorced white women.

Background

Divorced women qualify for Social Security benefits as retired workers, divorced spouses, or surviving divorced spouses. They can also receive widow benefits based on a prior marriage that ended in widowhood. Retired-worker benefits are computed by wage indexing annual earnings over a divorced woman's working life, then calculating her average indexed monthly earnings to determine her primary insurance amount (PIA)—the benefit payable at the full retirement age, which currently is 66. Individuals with 40 or more quarters of coverage over their work lives are fully insured and may receive retired-worker benefits.

For a divorced woman, the Social Security Administration (SSA) computes auxiliary benefits for each eligible marriage among the previous marriages she reports. Any person with a previous marriage that ended in divorce is eligible for auxiliary benefits if the ex-spouse was fully insured for Social Security benefits and the marriage lasted at least 10 years. Any person with a previous marriage that ended in widowhood is also eligible for auxiliary benefits if the deceased spouse was a fully insured worker. The 10-year marriage requirement does not apply to widow(er)s.1 Auxiliary benefits reflect the earnings history of the ex-husband, deceased ex-husband, or deceased husband from each marriage. If a woman's ex-husband is alive when she claims Social Security benefits, her auxiliary benefit (also known as divorced spouse benefit) is effectively equal to one-half his PIA.2 If a divorced woman's ex-husband is dead when she claims benefits, her auxiliary benefit (also known as a surviving divorced spouse benefit) is effectively equal to the deceased ex-husband's full PIA. Likewise, if a woman is widowed, her auxiliary benefit (also known as a widow benefit) is effectively equal to the deceased husband's full PIA.

After computing an auxiliary benefit for each eligible marriage, SSA selects the highest auxiliary benefit and compares it with the divorced woman's own retired-worker benefit. If she is not entitled to a retired-worker benefit, she receives the full auxiliary benefit as a divorced spouse, surviving divorced spouse, or widow beneficiary. If she is entitled to a retired-worker benefit that is less than the auxiliary benefit, she is "dually entitled" and SSA supplements her retired-worker benefit with the difference between her retired-worker benefit and the full auxiliary benefit to which she is entitled. Finally, if she is entitled to a retired-worker benefit that exceeds the auxiliary benefit, she receives only the retired-worker benefit.

Thus, a divorced woman's Social Security retirement benefit depends not only on her own earnings history, but also to a large extent on her marital history and the earnings histories of her previous spouses. Furthermore, a divorced woman with multiple marriages could receive an auxiliary benefit from any of her former spouses. Although she describes herself as divorced, at retirement she may receive a divorced spouse benefit, surviving divorced spouse benefit, or widow benefit from Social Security. In cases where none of her marriages ended in widowhood or in divorce after 10 years, a divorced woman will be ineligible for any auxiliary benefits.

Methods

We assess the retirement prospects of divorced women using the latest version of SSA's Modeling Income in the Near Term, version 6 (MINT6). MINT6 bases its projections on data from the 2001 and 2004 SIPP matched to Social Security administrative earnings and benefit data through 2008. For individuals born from 1926 through 1975, MINT6 projects each person's marital changes, mortality, entry to and exit from Social Security Disability Insurance rolls, and age of first receipt of Social Security retirement benefits.3 It also projects family income including Social Security benefits, pension income, asset income, earnings, Supplemental Security Income (SSI), income from coresident household members, and imputed rental income.4 Asset balances in retirement accounts and financial assets outside of retirement accounts in MINT6's starting SIPP sample are adjusted to align with distributions in the 2004 Survey of Consumer Finances.5

MINT6 is ideal for this analysis because it directly measures the experiences of survey respondents as of the early 2000s, thus accounting for the first half of the working lives of the 1946–1965 birth cohorts and first third of the working lives of those born 1966–1975. MINT6 projects their income and characteristics into the future, adjusting for expected demographic and socioeconomic changes. MINT6 also accounts for major changes in the growth of economy-wide real earnings, the distribution of earnings both between and within birth cohorts, and the composition of the retiree population. All these factors will affect the retirement incomes of future retirees.

We analyze the characteristics, work experiences, Social Security benefits, and total income of non-Hispanic white, non-Hispanic black, and Hispanic divorced women at age 70. Our analysis excludes divorced women who are projected to ever receive Social Security Disability Insurance. Because of the legislated increase in the full retirement age, the increase in the delayed retirement credit, the elimination of the retirement earnings test after attaining the full retirement age, and changes in pension and health insurance incentives, older adults are increasingly likely to work into their late 60s. Given these trends, we report total income at age 70 to represent the characteristics and economic well-being of those who have actually retired. Our analysis is limited to divorced women born between 1946 and 1975. These women represent the baby boom (born 1946–1965) and generation X (born 1966–1975) cohorts.6 We report all income projections in 2011 price-adjusted dollars.

Results

All model results include breakdowns by race and ethnicity. We begin by describing the projected marital status at age 70 of all women born from 1946 through 1975 to assess the size of the future population of divorced women. All subsequent analyses are limited to divorced women at age 70. We compare their demographic characteristics and marriage durations. We then describe their work and earnings histories to better understand their projected Social Security and retirement incomes. Next, we examine their Social Security benefit type at age 70, given their earnings histories and those of all prior spouses. A discussion of their projected total income, including an analysis of all major sources of retirement income, follows. Finally, we use multivariate regression analysis to examine the effect of benefit type, work history, and other economic and demographic characteristics on divorced women's projected retirement income.

Marital Status of Retired Women

MINT6 projects that women are more likely to be married at age 70 than they are to be divorced, never married, or widowed, regardless of race and ethnicity (Table 1). Still, close to one in five white women, one in five black women, and one in six Hispanic women are expected to be divorced at age 70.

Table 1. Projected marital status of women born 1946–1975 at age 70, by race and ethnicity (percent)
Marital status White (non-Hispanic) Black (non-Hispanic) Hispanic
Mean Standard deviation Mean Standard deviation Mean Standard deviation
All 100   100   100  
Divorced 21 41 22 42 16*** 37
Never Married 6 24 25*** 43 7 25
Married 61 49 39*** 49 59** 49
Widowed 12 32 15*** 35 19*** 39
Observations 16,800 2,499 2,466
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
Rounded components of percentage distributions do not necessarily sum to 100.
** = Difference from whites is statistically significant at the 5% level.
*** = Difference from whites is statistically significant at the 1% level.

Characteristics of Retired Divorced Women

Table 2 compares characteristics of divorced women by race and ethnicity. Among divorced women, MINT6 projects an average age at first Social Security benefit receipt of 64, regardless of race and ethnicity. For those who claim benefits before reaching full retirement age, SSA permanently reduces the benefit amount to offset the longer period over which benefits will be paid.

Table 2. Projected characteristics of women born 1946–1975 who are divorced at age 70, by race and ethnicity
Marital status White (non-Hispanic) Black (non-Hispanic) Hispanic
Mean Standard deviation Mean Standard deviation Mean Standard deviation
Average age at first receipt of Social Security benefits (years) 64 2 64 2 64 2
Marital history
Average age at first marriage (years) 23 6 25*** 7 24* 6
One marriage (%) 62 49 76*** 43 76*** 43
Two marriages (%) 28 45 19*** 39 21*** 41
Three or more marriages (%) 11 31 5*** 23 3*** 18
Any marriage lasted 10 years (%) 71 46 60*** 49 68 47
Education
High school dropout (%) 3 16 5*** 22 20*** 40
High school graduate (%) 66 47 72*** 45 65 48
College graduate (%) 32 47 23*** 42 15*** 36
Immigration status
Native-born (%) 92 27 88*** 33 51*** 50
Foreign-born (%) 8 27 12*** 33 49*** 50
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
Rounded components of percentage distributions do not necessarily sum to 100.
* = Difference from whites is statistically significant at the 10% level.
*** = Difference from whites is statistically significant at the 1% level.

The projected marital histories of divorced women differ significantly by race and ethnicity. Marital histories, in combination with earnings histories, are important for determining Social Security benefit eligibility and amounts. Compared with white women, black and Hispanic women will wait slightly longer before marrying for the first time. The projected average age at first marriage is 23 for white women, 25 for black women, and 24 for Hispanic women. Although a large percentage of women will marry more than once before age 70, divorced white women are dramatically more likely than other divorced women to have multiple marriages. Thirty-nine percent of divorced white women will marry two or more times, compared with only 24 percent of both black and Hispanic divorced women. Finally, marriage duration also differs significantly by race and ethnicity. Divorced white and Hispanic women are more likely than divorced black women to have ever had a marriage last 10 or more years—compare 71 percent of white women and 68 percent of Hispanic women with 60 percent of black women. So, although black women will be slightly more likely than white and Hispanic women to be divorced at retirement, they will be significantly less likely to qualify for auxiliary Social Security benefits because they do not meet the 10-year marriage requirement.

Among divorced women at age 70, whites are less likely than blacks and Hispanics to have not finished high school and they are more likely to have completed college. Only 3 percent of divorced white women do not have a high school education compared with 5 percent of divorced black women and 20 percent of divorced Hispanic women. By contrast, 32 percent of divorced white women, 23 percent of divorced black women, and 15 percent of divorced Hispanic women have college degrees.

Finally, native-born adults tend to have more educational and career opportunities than immigrants, and because they have lived in the United States their entire lives, native-born adults are more likely to have long work histories that entitle them to Social Security benefits. Because of high immigration rates in the mid- to late 1990s, especially among Hispanics (Passel and Suro 2005), nearly one-half of divorced Hispanic women are foreign-born, compared with only 8 percent of divorced white women and 12 percent of divorced black women.

Work and Earnings Histories

Social Security, pensions, and retirement accounts all depend directly on an individual's own work and earnings history. Table 3 examines whether divorced women's work histories differ by race and ethnicity. MINT6 projects that work histories will be very similar for whites and blacks, but very different for Hispanics. Perhaps related to the large share of immigrants among them, only about one-half of Hispanic women will work 35 or more years between ages 22 and 70 in Social Security-covered jobs, compared with 69 percent of both white and black women. Over their lifetimes, divorced Hispanic women will average only 31 work years, while divorced white and black women will average 37 and 36 work years, respectively.

Table 3. Projected labor force experience and lifetime earnings of women born 1946–1975 who are divorced at age 70, by race and ethnicity
Marital status White (non-Hispanic) Black (non-Hispanic) Hispanic
Mean Standard deviation Mean Standard deviation Mean Standard deviation
Years of work a (%)
10 or fewer 4 18 4 19 10*** 30
11–34 28 45 27 45 39*** 49
35 or more 69 46 69 46 51*** 50
Average years of work 37 11 36 11 31*** 13
Lifetime average annual earnings b
Own ($) 50,000 43,000 46,000* 41,000 37,000*** 29,000
Shared with spouse c ($) 58,000 41,000 48,000*** 39,000 41,000*** 28,000
Ratio of shared-to-own earnings 1.16 0.97 1.04 0.95 1.11 0.97
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
Rounded components of percentage distributions do not necessarily sum to 100.
* = Difference from whites is statistically significant at the 10% level.
*** = Difference from whites is statistically significant at the 1% level.
a. Number of years with positive earnings from ages 22 through 70.
b. Reflects the average of highest 35 years of wage-indexed earnings from age 22 to 70.
c. Computed by assigning each woman half the total earnings of the couple in the years when the she is married and her own earnings in years she is not married.

In contrast to work histories, earnings histories differ significantly by race and ethnicity. We analyze two measures of lifetime earnings. Both measures average the highest 35 years of wage-indexed earnings between ages 22 and 70. The first measure, own lifetime earnings, counts only the divorced woman's earnings. The second measure, shared lifetime earnings, assigns one-half of the couple's total earnings in the years she is married and her own earnings in the years she is single. Among divorced women, own lifetime earnings average $50,000 per year (in 2011 dollars) for whites—9 percent higher than $46,000 for blacks and 35 percent higher than $37,000 for Hispanics. Accounting for the earnings of former husbands increases the differences in earnings by race and ethnicity. Shared lifetime earnings average $58,000 per year for white women—21 percent higher than $48,000 for blacks and 41 percent higher than $41,000 for Hispanics. The difference between shared and own lifetime earnings among divorced women is 16 percent for whites and 11 percent for Hispanics, but only 4 percent for blacks. Because Social Security benefits are based on an individual's own earnings history and the earnings histories of current or former spouses, this finding suggests that the Social Security benefits of divorced black women will depend mostly on their own earnings histories and very little on their ex-husbands' earnings histories.

Social Security Benefit Status

Women's labor force participation and earnings have increased dramatically since the 1950s, and the gap between men's and women's earnings has steadily declined (Blau and Kahn 2007; Goldin 2006; SSA 2011). Consequently, many divorced women will be eligible for retired-worker benefits based on their own lifetime earnings. Table 4 reports how Social Security benefit status varies by race and ethnicity for divorced women. MINT6 projects that 96 percent of white, 92 percent of black, and 86 percent of Hispanic divorced women will receive retired-worker benefits (combining retired-worker only and dually entitled beneficiaries) at age 70. Black women are significantly more likely than white women to receive only retired-worker benefits at age 70. Among divorced women, 72 percent of blacks will receive only retired-worker benefits, compared with 65 percent of whites and 68 percent of Hispanics. However, more than one-half of divorced black women (37 percent of 72 percent) who receive only retired-worker benefits will do so solely because they failed to meet the 10-year marriage requirement. By contrast, more than one-half of divorced white and Hispanic women will receive only retired-worker benefits despite having met the a 10-year marriage requirement because the benefits, based on their own lifetime earnings, are higher than any auxiliary benefits they would be entitled to based on their former husbands' lifetime earnings.

Table 4. Projected percentage distribution of women born 1946–1975 who are divorced at age 70, by type of Social Security benefit received, and race and ethnicity
Type of benefit White (non-Hispanic) Black (non-Hispanic) Hispanic
Mean Standard deviation Mean Standard deviation Mean Standard deviation
All 100   100   100  
Retired worker only 65 48 72*** 45 68 47
Qualifying (10-year) marriage 37 48 35 48 42* 49
No qualifying marriage 28 45 37*** 48 27 44
Dually entitled 31 46 20*** 40 18*** 38
Spouse 7 25 2*** 15 4*** 19
Surviving spouse or widow 24 43 18*** 38 14*** 35
Auxiliary benefit only 2 15 4 19 3 16
Spouse 1 12 2 14 2 13
Surviving spouse or widow 1 10 2 13 1 9
Nonbeneficiary 3 16 4* 20 11*** 31
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
Rounded components of percentage distributions do not necessarily sum to 100 or to category subtotals.
* = Difference from whites is statistically significant at the 10% level.
*** = Difference from whites is statistically significant at the 1% level.

Although the gap between men's and women's earnings has been narrowing, most men still earn more than women. As such, a sizable share of divorced women will be eligible for auxiliary benefits based on their former husbands' earnings histories. Social Security will pay auxiliary benefits at age 70 to 33 percent of whites, 24 percent of blacks, and 21 percent of Hispanics. Racial and ethnic differences among auxiliary beneficiaries can be explained in part by differences in the correlation of earnings between divorced women and their ex-spouses (see ratio of shared to own lifetime earnings in Table 3), and in part by differences in the proportion of divorced women meeting the 10-year marriage requirement (see Table 2). A small percentage of divorced women (2 to 4 percent), regardless of race or ethnicity, will receive only auxiliary benefits at age 70 because they do not qualify for benefits on their own work histories (although they do qualify on their former husbands' work histories).

Most divorced white and black women will qualify for Social Security at age 70 based on their own or their former husband's work histories, but a significant number of divorced Hispanic women will not. Given their comparatively brief work histories, it is not surprising that more than one in ten divorced Hispanic women (many of whom are immigrants) will not collect Social Security benefits at age 70 (Table 4). By contrast, only 3 percent of divorced white women and 4 percent of divorced black women will not receive benefits.

Sources of Retirement Income

Table 5 presents the projected total income available for divorced women at age 70. Total income comprises income from assets, earnings, SSI payments, income from imputed rent, Social Security benefits, defined benefit pensions, and income from retirement accounts.

Table 5. Projected sources of total income for women born 1946–1975 who are divorced at age 70, by race and ethnicity
Type of benefit White (non-Hispanic) Black (non-Hispanic) Hispanic
Mean Standard deviation Mean Standard deviation Mean Standard deviation
  Percentage with income
Asset income 87 33 73*** 45 81*** 39
Earnings 29 46 27 44 19*** 39
SSI payments 2 15 5*** 22 9*** 28
Imputed rental income 71 45 61*** 49 54*** 50
Social Security benefits 98 16 96* 20 89*** 31
Defined benefit pensions 30 46 33* 47 21*** 41
Retirement accounts 72 45 66*** 48 56*** 50
  Mean total income ($)
Total 74,000 211,000 45,000*** 53,000 48,000*** 108,000
Asset income 29,000 204,000 8,000*** 40,000 14,000*** 98,000
Earnings 8,000 16,000 6,000** 14,000 5,000*** 12,000
SSI payments a 1,000 a*** 1,000 1,000*** 2,000
Imputed rental income 7,000 17,000 4,000*** 8,000 5,000*** 12,000
Social Security benefits 18,000 7,000 16,000*** 7,000 14,000 8,000
Defined benefit pensions 3,000 9,000 5,000** 12,000 3,000 9,000
Retirement accounts 9,000 19,000 7,000*** 16,000 8,000 34,000
  Share of mean total income (percentage distribution)
Total 100 . . . 100 . . . 100 . . .
Asset income 39 . . . 17 . . . 28 . . .
Earnings 10 . . . 14 . . . 10 . . .
SSI payments 0 . . . 1 . . . 1 . . .
Imputed rental income 9 . . . 8 . . . 10 . . .
Social Security benefits 24 . . . 35 . . . 28 . . .
Defined benefit pensions 4 . . . 10 . . . 6 . . .
Retirement accounts 12 . . . 15 . . . 16 . . .
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
Rounded components of percentage distributions do not necessarily sum to 100.
. . . = not applicable.
* = Difference from whites is statistically significant at the 10% level.
** = Difference from whites is statistically significant at the 5% level.
*** = Difference from whites is statistically significant at the 1% level.
a. Less than $500.

Among divorced women, whites are most likely to have earnings, imputed rental income, Social Security benefits, and retirement accounts at age 70, while Hispanics are least likely to, and blacks fall in the middle. For example, 29 percent of white women will have at least some earnings at age 70, compared with 27 percent of black women and only 19 percent of Hispanic women. Seventy-one percent of white women will be homeowners with imputed rental income, compared with only 61 percent of black women and 54 percent of Hispanic women. Seventy-two percent of white women will have retirement accounts, versus 66 percent of black women and 56 percent of Hispanic women. It is not surprising that 9 percent of Hispanic women will receive means-tested SSI benefits, compared with only 5 percent of black women and 2 percent of white women. What is somewhat surprising is that a relatively large share of Hispanic women will have income from assets, which includes the annuitized value of vehicles, other real estate, and farm and business equity; stock, mutual fund, and bond values; and checking, saving, money market, and certificate of deposit account balances, less unsecured debt. Among divorced women, 87 percent of whites, 81 percent of Hispanics, and only 73 percent of blacks will have at least one of these assets at age 70.

MINT6 projects average total income at age 70 to be at least 50 percent higher for divorced white women than for divorced black and Hispanic women; compare $74,000 for whites with $45,000 for blacks and $48,000 for Hispanics.7 Table 5's middle panel presents mean income by source, and the bottom panel shows the shares of mean income attributable to each source. Among divorced white women, on average, $29,000 (39 percent) of total income is projected to come from asset income, $8,000 (10 percent) from earnings, $7,000 (9 percent) from imputed rental income, $18,000 (24 percent) from Social Security benefits, $3,000 (4 percent) from defined benefit pensions, and $9,000 (12 percent) from retirement accounts.8

Among divorced black women, income from assets accounts for only $8,000, or 17 percent, of total income. Although their projected income from earnings ($6,000), Social Security benefits ($16,000), and retirement accounts ($7,000) are slightly lower than those projected for white women, these income sources comprise significantly larger shares of total income for divorced black women—at 14 percent for earnings, 35 percent for Social Security benefits, and 15 percent for retirement accounts. Finally, divorced black women are projected to receive an average of $5,000, or 10 percent of their total income, from defined benefit pensions.

Divorced Hispanic women are projected to have income from assets averaging $14,000, accounting for 28 percent of total income. Social Security benefits are projected to match assets as their largest income source, followed by income from retirement accounts ($8,000, or 16 percent of total income), then earnings and imputed rental income (each accounting for $5,000 or 10 percent of total income), and finally defined benefit pensions, which make up $3,000 or 6 percent of total income.

Despite having higher average incomes than divorced black women, divorced Hispanic women are more likely to be poor. Among divorced women born 1946–1975, 15 percent of Hispanics are likely to be poor at age 70, compared with 13 percent of blacks and 7 percent of whites (Chart 1).

Chart 1.
Projected poverty rates at age 70 for divorced women born 1946–1975, by race and ethnicity
Bar chart. White (non-Hispanic): 7 percent. Black (non-Hispanic): 13 percent*. Hispanic: 15 percent*.
SOURCE: Authors' calculations using MINT6.
NOTES: Total income is adjusted for household size using US poverty thresholds for persons aged 65 or older.
* = Difference from whites is statistically significant at the 10% level.

The Effect of Demographic and Socioeconomic Characteristics on Income

Next, we consider the combined effect of the variables described above on divorced women's retirement income. To do this we estimate a regression of the log of total income at age 70 by race and ethnicity. The results are presented in Table 6.9 Most of the variables have the expected correlation with retirement income. Delaying Social Security claiming and working 35 or more years is associated with higher-than-average retirement income for divorced women, regardless of race or ethnicity. Conversely, working 10 or fewer years and having no pension or asset income are related to lower-than-average retirement income for all divorced women.

Table 6. Marginal effects from regression of log income for women born 1946–1975 who are divorced at age 70, by race and ethnicity
Type of benefit White (non-Hispanic) Black (non-Hispanic) Hispanic
Age at Social Security benefit receipt  1.87***  1.86  4.62*
Age at Social Security benefit receipt squared -0.01*** -0.01 -0.01*
Marital history
Age at first marriage  0.04***  0.03  0.02
Age at first marriage squared  0.00***  0.00*  0.00
Two marriages -0.01  0.00 -0.10
Three or more marriages -0.04  0.01  0.06
Education
No high school diploma -0.29*** -0.17  0.03†††
High school graduate (omitted)
College degree  0.63***  0.58**  0.57***
Foreign born  0.12** -0.12†† -0.03
Years of work
10 or fewer -0.19** -0.34** -0.32
11 to 34 (omitted)
35 or more  0.42***  0.66***†  0.64***
Beneficiary status
Nonbeneficiary -0.37*** -0.24 -0.51***
Auxiliary benefits only  0.15 -0.04  0.10
Dually entitled  0.05 -0.01  0.14
Dually entitled via ex-spouse -0.02 -0.13 -0.43**††
Retired-worker only (omitted)
Retired worker with no 10-year marriages -0.15*** -0.07 -0.11
No pension or retirement account -0.37*** -0.29*** -0.36***
No asset income -0.39*** -0.25***††† -0.25***†
Constant -1.00 -1.00 -1.00
R-squared 0.373 0.482 0.389
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
* = Statistically significant at the 10% level.
** = Statistically significant at the 5% level.
*** = Statistically significant at the 1% level.
† = Difference from whites is statistically significant at the 10% level.
†† = Difference from whites is statistically significant at the 5% level.
††† = Difference from whites is statistically significant at the 1% level.

For divorced white women, the factors contributing most to higher-than-average retirement incomes are educational attainment and work experience. For example, those with college degrees have 63 percent more income than those with only high school diplomas, and those who worked at least 35 years have 42 percent more income than those who worked 11–34 years.

For divorced black women, college education and work experience are the only factors that are positively correlated with retirement income. Those with college degrees have 58 percent more income at age 70 than high school graduates, and those who worked at least 35 years have 66 percent more income at age 70 than those who worked 11–34 years.

Educational attainment and work experience are also correlated with higher-than-average retirement incomes among divorced Hispanic women. Average incomes are 64 percent higher for those with 35 or more years of work experience than they are for those with 11–34 years of work experience, and they are 57 percent higher for college graduates than for high school graduates.

For divorced white women, the factors contributing most to lower-than-average retirement incomes are beneficiary status, assets, pensions, and educational attainment. Average retirement incomes are 37 percent lower for nonbeneficiaries than for retired-worker beneficiaries, 39 percent lower for those without assets than for those with assets, 37 percent lower for those without defined benefit pensions or retirement accounts than for those with them, and 29 percent lower for those without high school diplomas than for those with them.

Many of these same factors are negatively correlated with retirement incomes among divorced black women—although to a much lesser degree. For black women, average retirement incomes are 34 percent lower for those who worked 10 or fewer years than for those who worked 11–34 years, 29 percent lower for those without pensions or retirement accounts, and 25 percent lower for those without assets.

Among divorced Hispanic women, the largest factor contributing to lower-than-average retirement incomes is beneficiary status. Compared with the retirement incomes of retired-worker beneficiaries, nonbeneficiaries' incomes are 51 percent lower and dually entitled spouse beneficiaries' incomes are 43 percent lower. In addition, average incomes at age 70 are 36 percent lower for those without pensions or retirement accounts and 25 percent lower for those without assets.

Conclusions

Historical demographic and marital trends suggest that minorities and divorced individuals will represent increasing shares of future retiree populations. This is a concern because minorities and the divorced have lower incomes and higher poverty rates in retirement than whites and other marital status groups.

Our microsimulation results show large differences in the characteristics, work experience, Social Security benefit status, and economic well-being of divorced women by race and ethnicity. MINT6 projects that on all measures, whites will be economically advantaged in retirement when compared with blacks and Hispanics. Although MINT6 projects that divorced Hispanic women will have slightly higher average incomes at age 70 than divorced black women, divorced Hispanic women are slightly more likely than divorced black women to be poor at age 70.

MINT6 projects that divorced women will continue to have high poverty rates in retirement, especially among minorities. Divorced women receive Social Security auxiliary benefits if their marriage lasted at least 10 years and their PIAs are less than one-half their ex-husbands' PIAs. However, minority divorced women are significantly less likely to have a qualifying marriage than white divorced women. Furthermore, the ex-husbands of minority divorced women are likely to have lower earnings than those of white divorced women. Both factors mean that white divorced women gain more from Social Security auxiliary benefits than minority divorced women.

Our analysis shows that more education, longer work careers, and access to pensions and savings significantly increase retirement incomes. Policy options that encourage more education and easier access to pensions can increase retirement security. Options that increase a woman's ability to enter or remain in the labor market, especially after divorce, can also significantly increase retirement incomes of divorced women. Working longer not only increases Social Security benefits, it also allows individuals more time to accumulate pensions and savings and reduces the number of years accumulated savings must support (Butrica, Smith, and Steuerle 2007).

Many of the factors correlated with higher retirement incomes for whites are also correlated with higher retirement incomes for blacks and Hispanics—namely, a college degree, strong labor force attachment, receipt of Social Security benefits, and having pensions, retirement accounts, or assets. However, because black and Hispanic women are less likely than white women to be college graduates, have strong work histories, receive Social Security benefits, or to own pensions, retirement accounts, or assets, their average retirement incomes are lower than those of white women.

Any Social Security reform Congress implements should recognize the continued economic vulnerability of divorced women. The current provisions for auxiliary benefits often fail to benefit many low-income divorced women. Options that reduce auxiliary benefits for higher-income spouses in place of higher worker benefits could increase Social Security benefits for many low-income divorced women without changing program outlays.

Appendix

Table A-1. Regression of log income for women born 1946–1975 who are divorced at age 70, by race and ethnicity
Type of benefit White (non-Hispanic) Black (non-Hispanic) Hispanic
Coefficient Standard error Coefficient Standard error Coefficient Standard error
Age at Social Security benefit receipt 1.053*** 0.336 1.050 0.707 1.727* 0.996
Age at Social Security benefit receipt squared -0.008*** 0.003 -0.008 0.005 -0.013* 0.008
Age at Social Security benefit receipt missing because not in labor force 34.674*** 10.839 36.119 22.902 58.175* 32.393
Marital history
Age at first marriage 0.039*** 0.012 0.032 0.021 0.018 0.038
Age at first marriage squared -0.001*** 0.000 -0.001* 0.000 0.000 0.001
Two marriages -0.007 0.030 -0.001 0.072 -0.107 0.100
Three or more marriages -0.038 0.043 0.012 0.126 0.055 0.232
Education
No high school diploma -0.349*** 0.082 -0.188 0.128 0.027††† 0.110
High school graduate (omitted)
College degree 0.487*** 0.029 0.459*** 0.070 0.451*** 0.121
Foreign born 0.110** 0.050 -0.128†† 0.088 -0.026 0.091
Years of work
10 or fewer -0.205** 0.102 -0.421** 0.193 -0.384 0.255
11 to 34 (omitted)
35 or more 0.349*** 0.032 0.505***† 0.073 0.496*** 0.100
Beneficiary status
Nonbeneficiary -0.463*** 0.141 -0.272 0.290 -0.719*** 0.251
Auxiliary benefits only 0.136 0.124 -0.042 0.218 0.093 0.328
Dually entitled 0.052 0.033 -0.008 0.080 0.133 0.123
Dually entitled via ex-spouse -0.025 0.055 -0.136 0.194 -0.568**†† 0.235
Retired-worker only (omitted)
Retired worker with no 10-year marriages -0.167*** 0.031 -0.068 0.064 -0.117 0.101
No pension or retirement account -0.455*** 0.032 -0.338*** 0.069 -0.439*** 0.094
No asset income -0.493*** 0.039 -0.285***††† 0.063 -0.284***† 0.106
Constant -25.209** 10.865 -26.728 22.961 -47.662 32.481
R-squared 0.373 0.482 0.389
Observations 3,549 581 403
SOURCE: Authors' calculations using MINT6.
NOTES: Sample excludes women projected ever to receive Disability Insurance benefits.
* = Statistically significant at the 10% level.
** = Statistically significant at the 5% level.
*** = Statistically significant at the 1% level.
† = Difference from whites is statistically significant at the 10% level.
†† = Difference from whites is statistically significant at the 5% level.
††† = Difference from whites is statistically significant at the 1% level.

Notes

1 Widow(er)s must have been married for at least 9 months to be eligible for widow(er) benefits. However, the 9-month marriage requirement is waived under certain conditions, such as when the spouse was "reasonably expected to live for 9 months" and death was accidental or in the line of duty.

2 We say "effectively" because Social Security benefits are reduced for early claiming and increased for delayed claiming.

3 MINT6 also projects outcomes for individuals born from 1976 through 2070 using a somewhat different approach from that used for the core cohorts born from 1926 through 1975. However, this analysis is only concerned with individuals born from 1946 through 1975.

4 We annuitize assets in MINT6 to represent the potential, rather than actual, income from assets since most retirees do not convert their financial assets into annuities. MINT6 takes the stock of wealth in nonpension, nonhousing assets and retirement accounts and (1) annually decays it based on age-wealth patterns in the SIPP to represent the spend-down of assets over retirement; and (2) converts assets into income by calculating the annuity a couple or individuals could buy if they annuitized 80 percent of their total wealth. Thus, asset income is derived from a series of annuity estimates based on a declining stock of wealth in retirement. Also, we calculate imputed rental income as a 3-percent real rate of return on home equity.

5 For more detailed information about the MINT model, see Smith and others (2010), Smith and others (2007), and Smith, Cashin, and Favreault (2005). Further sources of information are available at http://www.socialsecurity.gov/policy/about/mint.html.

6 The baby boom cohort is typically represented as those born between 1946 through 1964. For analytical purposes, however, we define the baby boom cohort as those born between 1946 and 1965.

7 Because of the positively skewed nature of assets, average asset values tend to be much higher than median asset values. Excluding asset income from total retirement income, racial and ethnic differences in average retirement incomes among divorced women would be much smaller. Although whites would still have the most income at age 70, Hispanics would now have the least (compare $45,000 for whites with $37,000 for blacks and $34,000 for Hispanics).

8 Because asset income in MINT6 is more comprehensive than in most surveys, its share of total income tends to be higher.

9 Table 6 shows the marginal effects. Appendix Table A-1 shows parameter estimates and standard errors for the ordinary least squares regression.

References

Ahlburg, Dennis A., and Carol J. De Vita. 1992. "New Realities of the American Family." Population Bulletin 47(2): 1–44.

Blau, Francine D., and Lawrence M. Kahn. 2007. "Changes in the Labor Supply Behavior of Married Women: 1980–2000." Journal of Labor Economics 25(3): 393–438.

Butrica, Barbara A., and Howard M. Iams. 2000. "Divorced Women at Retirement: Projections of Economic Well-Being in the Near Future." Social Security Bulletin 63(3): 3–12.

———. 2003. "The Impact of Minority Group Status on the Projected Retirement Income of Divorced Women in the Baby Boom Cohort." Journal of Women & Aging 15(2/3): 67–88.

Butrica, Barbara A., and Karen E. Smith. 2012. "The Retirement Prospects of Divorced Women." Social Security Bulletin 72(1): 11–22.

Butrica, Barbara, Karen E. Smith, and Eugene Steuerle. 2007. "Working for a Good Retirement." In Government Spending on the Elderly, edited by Dimitri B. Papadimitriou, 141–174. New York, NY: Palgrave Macmillan.

Census Bureau. 1995. Statistical Abstract of the United States: 1995. Washington, DC: Government Printing Office.

———. 2011. Statistical Abstract of the United States: 2011. Washington, DC: Government Printing Office.

DaVanzo, Julie, and M. Omar Rahman. 1993. "American Families: Trends and Correlates." Population Index 59(3): 350–386.

Goldin, Claudia. 2006. "The Quiet Revolution that Transformed Women's Employment, Education and Family." American Economic Review 96(2): 1–21.

Goldstein, Joshua R. 1999. "The Leveling of Divorce in the United States." Demography 36(3): 409–414.

Goldstein, Joshua, and Catherine Kenney. 2001. "Marriage Delayed or Marriage Forgone? New Cohort Forecasts of First Marriage for U.S. Women." American Sociological Review 66 (4): 506–519.

Harrington Meyer, Madonna, Douglas A. Wolf, and Christine L. Himes. 2005. "Linking Benefits to Marital Status: Race and Diminishing Access to Social Security Spouse and Widow Benefits in the U.S." Feminist Economics 11(2): 145–162.

———. 2006. "Declining Eligibility for Social Security Spouse and Widow Benefits in the United States?" Research on Aging 28(2): 240–260.

Kreider, Rose M., and Renee Ellis. 2011. "Number, Timing, and Duration of Marriages and Divorces: 2009." Current Population Reports P70-125. Washington, DC: Census Bureau.

[NCHS] National Center for Health Statistics. 1991. "Annual Summary of Births, Marriages, Divorces, and Deaths: United States, 1990." Monthly Vital Statistics Report 39(13). Hyattsville, MD: Public Health Service.

Norton, Arthur J., and Louisa F. Miller. 1992. "Marriage, Divorce, and Remarriage in the 1990's." Current Population Reports: Special Studies. Series P-23, No. 180. Washington, DC: Census Bureau.

Passel, Jeffrey, and Roberto Suro. 2005. "Rise, Peak, and Decline: Trends in U.S. Immigration 1992–2004." Washington, DC: Pew Hispanic Center.

Smith, Karen E., David B. Cashin, and Melissa M. Favreault. 2005. Modeling Income in the Near Term 4. Washington, DC: Urban Institute Press. http://www.urban.org/UploadedPDF/411191_MINT4.pdf.

Smith, Karen E., Melissa M. Favreault, Barbara A. Butrica, and Philip Issa. 2010. Modeling Income in the Near Term Version 6. Washington, DC: Urban Institute Press. http://www.urban.org/url.cfm?ID=412479.

Smith, Karen E, Melissa M. Favreault, Caroline Ratcliffe, Barbara Butrica, Eric Toder, and Jon M. Bakija. 2007. Modeling Income in the Near Term 5. Washington, DC: Urban Institute Press.

[SSA] Social Security Administration. 2010. Income of the Population 55 or Older, 2008. Washington, DC: SSA.

———. 2011. Annual Statistical Supplement to the Social Security Bulletin, 2010. Washington, DC: SSA.

Stevenson, Betsey, and Justin Wolfers. 2007. "Marriage and Divorce: Changes and Their Driving Forces." Journal of Economic Perspectives 21(2): 27–52.

Sweeney, Megan M., and Julie A. Phillips. 2004. "Understanding Racial Differences in Marital Disruption: Recent Trends and Explanations." Journal of Marriage and Family 66(3): 639–650.

Tamborini, Christopher R., Howard M. Iams, and Kevin Whitman. 2009. "Marital History, Race, and Social Security Spouse and Widow Benefit Eligibility in the United States." Research on Aging 31(5): 577–605.

Tamborini, Christopher R., and Kevin Whitman. 2007. "Women, Marriage, and Social Security Benefits Revisited." Social Security Bulletin 67(4): 1–20.

Tejada-Vera, Betzaida, and Paul D. Sutton. 2010. "Births, Marriages, Divorces, and Deaths: Provisional Data for 2009." National Vital Statistics Reports 58(25). Hyattsville, MD: National Center for Health Statistics.