This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

Crediting Care or Marriage? Reforming Social Security Family Benefits

LBJ School of Public Affairs, University of Texas, Austin.

Address correspondence to Pamela Herd, University of Wisconsin, Madison, Department of Sociology, 1180 Observatory Drive, Madison, WI 53706. E-mail: pherd{at}Lafollette.wisc.edu

	Abstract

TOP Abstract Methods Results Discussion References

 
Objectives. For more than 20 years policy advocates and policymakers have argued that Social Security should reward women for raising children. Current family benefits, which only benefit women who marry, are thought to be outdated and unable to protect the neediest women. Thus, would Black and poor women fare better if family benefits were linked to parenthood, as opposed to marriage? I examined three care credit proposals that reflect the most common proposals put forth in the United States and the most common designs in other countries.

Methods. I used the 1992 Health and Retirement Study and the Current Population Survey to create a policy simulation that estimates how women reaching age 62 from 2020 to 2030 would be affected by care credits.

Results. Black and poor women fared best with benefits linked to parenthood. The specific proposal allowed parents, from the 35 earnings years used to calculate their benefit, to substitute $15,000 for up to 9 earnings' years that fell below this level.

Discussion. The poorest women fare better with family benefits linked to parenthood instead of marital status. Moreover, they fare best when working women can benefit from care credits, but the care credit's value is not linked to earnings.

Though often seen as a program that benefits workers, 40% of Social Security beneficiaries are the wives and widows of workers. While Social Security's fiscal problems and privatization have dominated the headlines, 25 years of research has documented that family benefits, which mainly benefit women who get married, stay married, and do no paid work, are outdated (Advisory Council on Social Security, 1979; Smeeding, Estes, & Glasse, 1999). Today, few families fit this breadwinner model.

The subject of this research is one oft proposed, but little studied, option to reform family benefits: increasing women's worker benefits by rewarding them for raising children. Monies spent on spousal benefits are diverted to subsidize the cost of these "care credits." Given that benefit adequacy is arguably Social Security's most important goal, I paid close attention to how rewarding parenting, as opposed to marrying, affects race and class inequality among women. Policy researchers have paid much attention to how an outdated benefit structure affects women and little attention to how it impacts different groups of women, despite the fact that poverty rates among Black and unmarried women are three times that of White married elderly women (Smeeding et al., 1999). Benefit adequacy is also a critical theoretical issue; how policies shape gender, race, and class stratification is the subject of considerable scholarly debate (Esping-Anderson, 1999). And while scholars have closely analyzed how the basis of entitlement for family benefits, be it marriage or parenthood, affects gender inequality, they have paid little attention to how it affects race and class inequality among women.

Should Family Benefits Reward Marriage or Parenthood? The Theoretical Debate
A key issue in welfare state research is how the welfare state shapes gender, race, and class inequality (Esping-Andersen, 1999). And the impact of family benefits on inequality has come under increasing scrutiny given the extent to which work and family life has changed during the past 30 years (Esping-Anderson; Orloff, 1993). One understudied question, however, is whether the basis of entitlement for family retirement benefits, be it marriage or parenthood, has differential impacts on inequality.

Currently, the wives and widows of retired workers are eligible for Social Security benefits. In fact, many industrialized countries give benefits to the spouses and/or widows of workers (Thomson & Carasso, 2002). Social Security entitlements linked to marriage ensure that married women, many of whom have worked little and for low earnings, have an adequate retirement benefit. Though feminists are nearly universally opposed to policies promoting marriage because they encourage women's economic dependence upon men, there is a feminist rationale for marriage-based benefits. Simply, married women contribute to men's earnings by taking care of all the reproductive labor that is necessary to survive. They cook, clean, and even organize their spouses' social lives. Feminists have long argued the economic contributions to households that women make in this regard (Secombe, 1974). Indeed, married men out earn single men (Loh, 1986). Thus, the state should recognize and reward these contributions. This may explain why feminist organizations, from the National Organization of Women to the Feminist Majority, have never challenged the eligibility link in Social Security between marital status and benefits.

Others argue that a better strategy for reducing gender inequality is to link family benefits to parenthood. All industrialized countries, except the United States, reward parenthood through their pension systems (Thomson & Carasso, 2002). Many argue that children, as opposed to marriage or labor force discrimination, are the key to understanding gender differences in earnings and ultimately income security in the 21^st Century (Crittenden, 2001; Waldfogel, 1997). Rewarding marriage would not subsidize the costs of raising children for all mothers. And the costs are high. Around 60% of women are not employed when they have children under a year old (Bachu & O'Connell, 2000). Women face a high price in foregone wages (Burggraf, 1998). Contrastingly, men face either no wage penalty or even an increase in their wages after having a child (Loh, 1986; Lundberg & Rose, 2000). Thus, the state should insure against the price of raising children by spreading the cost from individual women to the larger society (Cancian & Oliker, 2000; Folbre, 1994; Knijn & Kremer, 1997; Lister, 1997). This general call is the main justification for including care credits in Social Security, which would reward women for raising children by supplementing their worker benefit (Crittenden).

But while close attention has been paid to how family benefits affect gender inequality, little attention has been paid to how they affect race and class inequality among women. However, as the discussion below will illustrate, linking benefits to parental status, as opposed to marital status, may have different ramifications on race and class inequality among women.

Should Family Benefits Reward Marriage or Parenthood? The Policy Debate
An oft repeated critique of Social Security family benefits is that they are outdated because they best benefit women who get married, stay married, and do no paid work. Individuals receive contributory benefits as workers or noncontributory family benefits as the spouses and survivors of workers. Contributory worker benefits are based on an individual's highest 35 earnings years. Noncontributory family benefits have no relationship to the recipient's work history. Around 98% of noncontributory family benefit recipients are women (Harrington Meyer, 1996). A wife's spousal benefit is worth half of her husband's worker benefit, while the survivor receives 100% of her late husband's original worker benefit. A divorced woman can receive spousal and survivor benefits if she was married for at least 10 years. Women eligible for both worker and spousal benefits receive the higher of the two.

The problem, policy activists and scholars emphasize, is that it is increasingly unlikely that women will get married, stay married, and do no paid work. From 1970 to 2000, the percentage of women married dropped from 60% to 52%, while the percentage of women divorced more than doubled, from 6% to 13%. In regards to work, between 1960 and 1997 the percentage of married mothers in the workforce rose from 28% to 71% (Castro, 1998). But women are not working enough or earning enough to justify simply eliminating marital benefits. Women's average annual earnings are around just 60% of men's.

Though most of the discussion about outdated family benefits has focused on how this will impact women, an outdated benefit design is particularly problematic for Black and poor women. Black and poor women are twice as likely to never marry as are White and wealthier women (Goldstein & Kenney, 2001). Those who do not marry cannot qualify for these benefits. Moreover, Black women have consistently been more likely to work than have White women. In 1950, 43% of Black women worked compared to 28% of White women (Smith & Ward, 1985), though by the 1990s the gap had largely closed (Council of Economic Advisors, 1998). Contributions workers make through payroll taxes subsidize noncontributory family benefits. Thus, working Black women have been subsidizing White women receiving spousal and widow benefits. Moreover, for married women, the higher her spouse's income the less likely she is to go back to work after having a child (Barrow, 1999). Thus, the way the system is currently designed, a single woman working in a low-wage job all throughout her life will likely end up with a lower benefit than a woman who has never been employed but is married to a wealthy man (Harrington Meyer, 1996).

But others argue that Black and poor women have fared quite well in the current system. The poverty rate among African Americans would be twice as high if individuals had no Social Security income (Porter, Larin, & Primus, 1999). Further, for those in the bottom two fifths of the income distribution, Social Security comprises 80% of their income compared to 18% for those in the top fifth (Porter et al.). This income is critical to their economic well-being. Thus, while there are problems with marital status benefits for Black and poor women, would they fare better with care credits? To answer this question, I evaluated three different care credit proposals, which are outlined in the following section.

Care Credit Proposals
The care credit proposals reflect the varying designs employed in other countries but also resemble specific proposals that have been put forth in the United States. Austria, Belgium, Canada, France, Germany, Japan, Norway, Sweden, Switzerland, and the United Kingdom all link retirement family benefits to parenthood. Two factors determine the design of care credits in these countries: eligibility (Should working parents have their retirement benefits supplemented or should only those who leave employment be compensated? For example, France only credits mothers who have years of no earnings, while Norway credits mothers with years of low earnings.); and how the care credit is valued (Countries employ two different approaches. One is to compensate women based upon their lost earnings. A lawyer receives a higher credit than a secretary. Alternately, the care work is valued independently. The lawyer and the secretary receive the same compensation.).

Thus, I analyzed three care credit proposals that reward parents for raising children. The proposals not only represent approaches common in other countries, they also represent the most common proposals in the United States put forth to date. Their shared trait is the elimination of spousal benefits, which require no employment history. Thus, they all require an earnings' history for eligibility, unlike spousal benefits. Those resources then offset the costs of care credits, which do require an employment history. The differences between the proposals center on who is eligible and how the credit is valued. Are parents who work eligible? Is the care credit's value determined independently or based on prior earnings?

Drop zeros
Parents can drop up to 9 no earnings years (up to 5 for one child and up to 9 for more than one child) from the 35 used to calculate their benefits (Advisory Council on Social Security, 1979; Crittenden, 2001). Spousal benefits are eliminated. Only parents who fully exit paid labor can benefit. The credit's value is based on earnings during years they were employed.

Substitute
Parents could substitute up to 9 years of earnings (up to 5 for one child and up to 9 for more than one child) with half of the median wage (Gore, 2000). This is the median wage for those 15 years or older working full time, which was about $15,000 in 2000. Thus, working parents can receive the credit, but they must have relatively low earnings. Obviously, the value of the care credit is independent of earnings.

Drop Low earnings
The Drop Low proposal, like the Substitute proposal, will also benefit parents whether they do or do not stay employed. But, unlike the Substitute proposal, it puts no limits on how much they work. Parents can drop 5 (for one child) to 9 (for those with more than one child) of their lowest earnings years within the 35 used to calculate their worker benefit (Crittenden, 2001). The value of the credit is then linked to earnings.

None of these proposals require that care credits be specifically tied to years where children are young. The rationale behind this is that the burdens of childrearing vary significantly between individuals. For many, time off is needed when children are very young. For others, behavioral problems in the teen years require more time at home for supervision.

Study Overview
I used the 1992 Health and Retirement Study (HRS) and the 2000 Current Population Survey (CPS) to create a policy simulation that estimates how women reaching age 62 from 2020 to 2030 would be affected by the implementation of three different care credit proposals, all of which entail the elimination of spousal noncontributory benefits. I examined how each proposal changes the overall distribution of noncontributory family benefits or the benefits not based on the individuals' earnings. For example, do women in low-asset households, as a group, receive a greater proportion of family noncontributory benefits under current rules or under the care credit reforms? I focused further analyses on how women's total individual Social Security benefits change. In particular, which groups of women are most likely to get benefit increases and decreases, and what are the sizes of those increases and decreases?

Data
The HRS is a longitudinal study of individuals born between 1931 and 1941. The survey, which was started in 1992, is conducted every other year. It includes data on demographics, physical health, housing, family structure, current and past employment, retirement plans, income, and insurance. It is the most extensive data collection of a nationally representative cohort of older people (Mitchell, Hammond, & Rappaport, 2000).

The initial sample in Wave 1 of the HRS was 5,426. Of these, 3,971 women consented to the use of their Social Security records, which are essential to calculate Social Security benefits. Another 544 married women were excluded because their spouses' lacked earnings' records, leaving a sample of 3,427 women. Weight was utilized to adjust missing observations.

Any analysis of Social Security family benefits must be based on couple data. Evaluations that do not use couple data are inaccurate because individuals can draw auxiliary benefits off of their spouse's or ex-spouse's earnings history. The HRS, with attached Social Security earnings data, is the most complete data currently available. Yet, there are two problems in using the HRS to construct Social Security benefits: missing information on husbands for formerly married women and missing earnings years up to age 62 because individuals in the sample are ages 51 to 61. The latter problem was more easily solved. Waves 2, 3, 4, and 5 of the HRS filled in most of the missing earnings years while the rest were imputed based on the approach Mitchell and colleagues (2000) used to create a predicted wealth file for the HRS.

The more difficult issue was missing earnings records for the former spouses of current women beneficiaries who had been divorced or widowed. To address this, I employed multiple imputation (Rubin, 1987). I imputed earnings histories, for 583 women, from randomly selected male records with similar characteristics to the women. Divorced women who had never had a marriage that lasted at least 10 years and presently widowed women receiving survivor benefits did not need imputed earnings records. While error was introduced into this analysis by imputations, there is no other data that provides more complete information. The Social Security Administration uses the same approach to conduct policy analyses. Multiple imputation allowed for the adjustment of coefficients and standard errors to reflect the introduced error.

The last stage of data preparation entailed a demographic simulation to project what women born between 1960 and 1970 would look like at age 62. The purpose of this nonparametric projection was to see how a younger cohort of women, with different demographic characteristics, would be affected by the proposals. The premise of the simulation was that the 1931–1941 cohort in the HRS included women with employment and family histories similar to the 1960–1970 cohort, it was just that the proportion of different groups of women varied. For example, there were divorced women in the HRS cohort, but there were a greater number of divorced women in the 1960–1970 cohort. Thus, I used the 2000 CPS to develop a series of weights that adjust the HRS cohort of women to resemble the 1960–1970 cohort. Extensive family and work histories in the HRS allowed for the reconstruction of these women's characteristics between the ages of 30 and 40, which was the age of the 1960–1970 cohort in the 2000 CPS. Although this was a simplified projection, the results were similar to more complicated and expensive simulations, such as the MINT model created by the Brookings Institution, the Urban Institute, and the Social Security Administration. For more information on all of these data adjustments and imputations see Herd (2002).

Variables
The dependent variable represented the effect of the policy change from the present structure, which included spousal benefits, to one of the proposed three care credit options including dropping zero earnings years (Drop Zeros), substituting years with half of the median wage (Substitute), or dropping low earnings years (Drop Low). This dependent variable was continuous, with both positive values (for "gainers") and negative values (for "losers"), but with a substantial number of zeros (for those benefits left unchanged by the alternative policies).

Table 1 provides the means and distributions for the covariates. They included demographic, family status, socioeconomic, and work history variables. Year of birth and race wee the two primary demographic variables. Race was a dummy variable, with White women as the reference. The sample was not large enough to allow for separate analyses of Black and Hispanic women.

Family status variables included whether women have children or not; those with children were the reference. I kept this as a simple dummy because I wanted to know how women with children, versus those without children, fared. Additional analyses that included dummies for the number of children did not meaningfully alter the findings. Coded as a dummy, the reference category for marital status was married women, compared to divorced, widowed, and never married women.

The work history variables addressed how the number and timing of zero earnings years affected who benefits under the care credit proposals. One set of dummy variables represented the number of zero earnings years out of 30 years of earnings from age 20 to age 50: 0–5, 6–10, 11–15, 16–20, 21–25, and 26–30. In order to account for the timing of those zero earnings years, another variable addressed whether zero earnings years occurred in a woman's 20, 30s, or 40s. The reference category was women who had zero earnings in all three decades (named 111). Other categories include women who had no zero earnings years in any of those decades (000), zero earnings only in their 20s (100), zero earnings in their 20s and 30s (110), zero earnings only in their 30s (010), and zero earnings in their 30s and 40s (011). No women fell into the two other possible earnings configurations (101) and (001).

	METHODS

TOP Abstract Methods Results Discussion References

 
The Three-Equation Model
While descriptive analyses will provide a significant level of information, regression analyses will isolate the key correlates associated with who would win and lose under of each of the care credit reforms, as well as the extent to which benefits would rise or fall for those whose benefits changed. The baseline for comparison was the respondents' present benefit. Under each proposal, each recipient's current benefit may be decreased, left unchanged, or raised. The extent of any such change can be measured in dollar terms. There will be a large number of beneficiaries with no benefit change, thus creating zeros. Therefore, I used an altered model from the two-part model, the three-equation model, to investigate correlates of benefit changes under each reform proposal (Manning, Newhouse, Duan, & Leibowitz, 1987). The first part of the two-part model captured the probability of having a benefit change. The second part captured the size of that change.

The first equation (first part) in the model was an ordered probit where covariates included demographic, socioeconomic, family status, and work history individual characteristics. The dependent variable was ordinal with categories corresponding to a decrease in benefit (y = 0), no change in benefit (y = 1), and an increase in benefit (y = 2). The second equation (second part) was a linear regression of individuals' increases in benefits given that they were in the increase category in the first equation. The third equation (second part) was a linear regression of individuals' decreases in benefits among those experiencing a decrease. The dependent variable in the latter two equations was the natural log of the benefit change.

	RESULTS

TOP Abstract Methods Results Discussion References

 
Overall Change in Benefit Distribution
Table 2 overviews the policy proposals. All of the results reflected outcomes for women reaching age 62 between 2020 and 2030. Figures 1 and 2 show what proportion of noncontributory benefits will go to poor versus wealthy women and Black versus White women under care credits as compared to the current rules. The outcome measure was a fraction where the denominator was the sum (in dollars) of noncontributory family benefits overall, and the numerator was the sum (in dollars) of noncontributory family benefits received by different groups of women. Noncontributory benefits supplemented the individual's worker benefit; they equaled the worker benefit subtracted from the family benefit (be it care credits or spousal benefits).

View larger version (35K): [in this window] [in a new window]   Figure 1. Distribution of noncontributory benefits by household asset level under current rules and care credit proposals

View larger version (21K): [in this window] [in a new window]   Figure 2. Distribution of noncontributory benefits by race under current rules and care credit proposals

Figure 2 presents the distribution of noncontributory family benefits by race. Black women comprised about 11% of the sample. With current spousal benefits, they received just 7% of the total sum of family benefits. All three proposals significantly improved the racial inequality built into the current system of family benefits. Drop Zeros, Substitute and Drop Low raised the proportion of noncontributory family benefits received by Black women to 10%, 11%, and 11%, respectively. These proposals were more advantageous for Black women because these benefits were more tightly linked to employment than spousal benefits. Black women have generally worked more than White women. Moreover, given Black women's lesser likelihood to be married, part of the advantage for Black women was de-linking family benefits from marital status.

Benefit Changes: Benefit Increases
The rest of the results reflected a traditional policy analysis where the outcome variable was the difference between individuals' benefits under the current rules compared to what their benefits would be under the proposed reform. Overall, the most women benefited from Substitute and Drop Low, while Black, poor, and unmarried women fared best under Substitute followed by Drop Low.

Table 3 displays percentages of individuals receiving benefit increases and decreases and the size of those increases and decreases. While just 27% of women received benefit increases under the Drop Zeros proposal, close to half of women received benefit increases under the Substitute and Drop Low proposals. This signals that most women in this cohort were working enough that only providing credits to women who entirely left employment for periods of time was not as advantageous to as many women as also providing credits to women who had more extensive employment histories. Average benefit increases were slightly more than 10% for all three proposals.

Unmarried women clearly fared better than married women under all of these proposals but fared best under Substitute and Drop Low. Almost double the percentages of divorced women received increases under the Substitute and Drop Low proposals compared to the Drop Zeros proposal. Among never married women 16% and 20% received increases under the Substitute and Drop Low, respectively, compared to 9% under Drop Zeros. More than one half (for Substitute and Drop Low) and one third (for Drop Zeros) of married women had benefit increases.

The size of those benefit increases was highest under Substitute compared to the other two proposals. Benefit improvements for divorced women were similar, around 10% for all three proposals. Never married women, however, had an average benefit improvement of 28% for Substitute, compared to 9% and 14% for Drop Low and Drop Zeros, respectively. Unmarried women, perhaps because this group was comprised disproportionately of Black women, fared best with policies that rewarded their childrearing, even if they were employed, and with policies that did not tie the value of the care credit to earnings.

Women in the bottom household asset quartile also fared best under the Substitute proposal. Compared to 43% of these women under Drop Low and 25% of these women under Drop Zeros, 46% received a benefit increase under Substitute. Moreover, under both the Drop Low and Drop Zeros proposals, higher percentages of women in the top asset households received benefit increases than in the bottom asset households. This was not the case with the Substitute proposal. Women in the lowest asset households had the highest benefits increases under Substitute at 17%, compared to 12% for Drop Low and 11% for Drop Zeros. Clearly, it was important for women in low-asset households that the value of the credit be determined independent from their earnings but also that they were not penalized for combing employment with childrearing.

Benefit Changes: Benefit Decreases
These benefit changes also led to benefit cuts for some women compared to their benefits under current rules. Overall, for those who received benefit cuts, the Substitute proposal did the least damage, followed by Drop Low, then Drop Zeroes. Because the value of Substitute was not linked to earnings, even those with meager earnings histories received a more generous benefit than under the other proposals, which did connect the value of the credit to earnings. Around one fifth of women received benefit decreases relative to their original benefit under these reforms. Black women, compared to White women, faced the fewest cuts under all of the proposals but fared best under Substitute. Just 12% of Black women had a benefit decrease of about 29% under Substitute. Contrastingly, about 13% of Black women received more than a 50% benefit reduction under Drop Low and Drop Zeros. Between 22% and 26% of White women had benefit decreases ranging between 30% for Substitute and 53% for Drop Low and Drop Zeros.

Both unmarried and married women fared better under Substitute than the other two proposals. Because it was the elimination of spousal benefits that led to benefit reductions, cuts were concentrated among eligible married and divorced women. Never married and widowed women did not receive reductions. Divorced women clearly fared best under Substitute where 14% received a 32% benefit reduction. Comparatively, 15% to 17% of divorced women received a 49% reduction under Drop Low and Drop Zeros. The benefit cut for about one third of married women was 30% under Substitute compared to more than 50% for Drop Low and Drop Zeros.

Finally, women in low-asset households fared best, in terms of benefit cuts, with Substitute as compared to Drop Low and Drop Zeros, though benefit cuts were more concentrated in high-asset households under all of the proposals. For 15% to 19% of women in low-asset households, the size of the reduction was 30% for Substitute compared to more than 50% for the other two proposals. This finding is of concern given that this is an economically vulnerable group. Ideally, policymakers would include provisions that would offset benefit decreases for this group with a policy change.

Benefit Changes: Replacement Rates
Given that the large majority of benefit cuts were concentrated among married women, benefit replacement rates in Table 4 give a clearer perspective on how these benefit changes affect married couples. Replacement rates, which are commonly used to examine the progressivity of any Social Security reform, are the monthly benefit divided by monthly average lifetime earnings or the percentage of average monthly lifetime earnings the Social Security benefit is replacing.

Benefit Changes: Regression Analyses
While the previous descriptive analyses are critical to any policy evaluation intended to understand how subgroups fare, Tables 3 and 4 do not control for more than one background factor at a time. Regression analyses clarify how very precise groups of women fare (by controlling for multiple individual characteristics) under each of the proposals.

The results from Table 5, 6, and 7 emphasize that those who had lower benefits under current rules and those with fewer years of zero earnings were most likely to benefit and the least likely to lose from Substitute, Drop Low, and Drop Zeroes respectively. These two variables were the strongest predictors of benefit gainers and losers with the switch to care credits, and the discussion that follows centers on these variables.

The ordered probit estimates in Table 5 are not intuitively interpretable, thus Table 6 presents predicted probabilities based on the ordered probit estimates. Given that married women were the ones most likely to lose under these proposals, Table 6 focuses specifically on them.

Compared to the other two proposals, the Substitute proposal best balanced benefiting mothers who remained employed while not penalizing them too harshly if they left employment. Under both Substitute and Drop Low married mothers few no earnings years had an 85% probability of benefit gain and around 5% probability of benefit loss. Under Drop Zero, they had just a 26% probability of benefit gain and a 35% probability of benefit loss. Those with many zero earnings had a 25%, 15%, and 7% probability of benefit gain under Substitute, Drop Low, and Drop Zero, respectively. Their probabilities for benefit cuts were around 70% for Drop Zero and Drop Low and 53% for Substitute.

Women with lower benefits under current rules, compared to women with higher benefits, fared best under the Substitute proposal compared to the Drop Zeros and Drop Low proposals. Neither Drop Zero nor Drop Low had meaningful differences in their impacts on benefit increases and decreases among women with $500 versus $1,000 current rule benefits. Around one in five married mothers would lose while around 41% and 60%, under Drop Zero and Drop Low, respectively, would gain. Under Substitute a married mother with a $500 benefit under current rules would have an 84% probability of gain and a 6% probability of loss. Those with a $1,000 current rule benefit would have a 69% probability of gain and 14% probability of loss.

Finally, how did different groups of women fare in terms of the size of their benefit increases and decreases, holding constant demographic factors? Table 7 presents the OLS results for those receiving benefit increases and benefit decreases. In terms of progressiveness, previous benefit size was the factor most strongly related to the size of benefit increases and decreases. The Substitute proposal was the most progressive of the three proposals: the higher an individual's original benefit, the smaller the increase and the higher the decrease under the reform. Contrastingly, under the Drop Low and Drop Zeros proposals an individual with higher current benefits ended up with a higher benefit increase. Household assets, contrastingly, were not significantly predictive of the size of benefit change. Table 7 also shows that among women with benefit increases and among those who worked, those who worked the least gained the most from the Drop Zeros proposal. Compared to women with less than 6 zero earnings years in their employment history, the benefit increase for a women with 16 to 20 zero earnings years was 122% higher under Drop Zeros compared to 50% and 28% under Substitute and Drop Low, respectively.

	DISCUSSION

TOP Abstract Methods Results Discussion References

 
For more than 20 years policy advocates and policymakers have argued for the implementation of care credits in Social Security. The United States stands out as the only western industrialized country that does not reward care within its state pension system. The results from this study clarify how implementing care credits would change the distribution of benefits among women beneficiaries reaching age 62 between 2020 and 2030.

Policy researchers and welfare state theorists have failed to consider how the basis of entitlement for family benefits, be it marriage or parenthood, affects race and class inequality among women. The results from this study show that linking family benefits to parenthood, as opposed to marriage, is most advantageous for poor and Black women. These women fared better both because they were less likely to be married but also more likely to have been employed. Care credits supplement the worker benefit, while spousal benefits have no link to women's earnings. Inequality among women is lessened when their benefits are based on their own work, be it in the workplace or raising children, as opposed to their spouses' work.

In terms of the practical policy implications, while all the proposals are more progressive than spousal benefits, the Substitute proposal is clearly the most progressive. Black and poor women fare best with the Substitute and Drop Low proposals, respectively. Both proposals allow mothers to benefit from combining their childrearing with employment, but the Substitute proposal values the care credit independently of women's earnings, while the Drop Low proposal values the care credit based on beneficiaries' earnings. The Drop Zeros proposal, though the most often proposed plan, benefits half as many women as do the Substitute and Drop Low proposals. Requiring full exits from employment to receive the credit, while simultaneously valuing that credit based on years where there are earnings, is a problematic mix for most women. While Drop Zero and Drop Low are marginally more progressive than spousal benefits, they are highly questionable solutions to the problem given their drawbacks.

One large negative outcome to these proposals, however, is the following: Resources are redistributed without increasing that pool of resources; thus, some are worse off under the new rules. The Substitute proposal, in particular, evens the playing field among mothers. The old rules favored married (and formerly married) mothers who were rarely employed (often White and wealthier women) at the expense of single mothers who worked (often poor and Black women). Nonetheless, while 45% of women would fare better under the new rules (the Substitute proposal) as compared to the old, 30% would fare worse. But declaring those changes should not be made because some will end up worse off than before condemns the state to perpetually reinforcing existing inequality.

	Acknowledgments

 
The research reported herein was supported (in part) by the Center for Retirement Research at Boston College pursuant to a grant from the U.S. Social Security Administration funded as part of the Retirement Research Consortium. The opinions and conclusions are solely those of the author and should not be construed as representing the opinions or policy of the Social Security Administration or any agency of the Federal Government or the Center for Retirement Research at Boston College.

The author would like to acknowledge Madonna Harrington Meyer, Doug Wolf, Tim Smeeding, and Don Moynihan for their helpful comments.

	Footnotes

 
Decision Editor: Charles F. Longino, Jr., PhD

Received for publication July 8, 2004. Accepted for publication July 11, 2005.

	References

TOP Abstract Methods Results Discussion References

 

• Advisory Council on Social Security. (1979). Social Security financing and benefits: Report of the 1979 advisory council. Washington, DC: Government Printing Office.
• Bachu, A., O'Connell, M. (2000). Trends in labor force participation and employment characteristics of married-couple families. Washington, DC: Census.
• Barrow, L. (1999). An analysis of women's return to work decisions. Economic Inquiry, 37,432-451.
• Burggraf, S. (1998). The feminine economy and economic man. New York: Basic Books.
• Cancian, F., Oliker, S. (2000). Caring and gender. Thousand Oaks, CA: Sage.
• Castro, I. (1998). Equal pay: A thirty-five year perspective. Washington, DC: U.S. Department of Labor.
• Council of Economic Advisors. (1998). Labor markets. In Changing America: Indicators of social and economic well-being by race and Hispanic origin (pp. 23–32). Washington, DC: Government Printing Office.
• Crittenden, A. (2001). The price of motherhood. New York: Metropolitan Books.
• Esping-Anderson, G. (1999). Social foundations of postindustrial economies. Oxford: Oxford University Press.
• Folbre, N. (1994). Children as public goods. American Economic Review, 84,86-90.
• Goldstein, J., Kenney, C. (2001). Marriage delayed or marriage forgone? American Sociological Review, 66,506-519.
• Gore, A. (2000). Social Security and retirement security. Retrieved January 15, 1999, from http://www.algore.2000.com.
• Harrington Meyer, M. (1996). Making claims as workers or wives: The distribution of Social Security benefits. American Sociological Review, 61,449-465.
• Herd, P. (2002). Crediting care, citizenship, or marriage: Women and Social Security reform. Unpublished doctoral dissertation, Syracuse University, Syracuse, New York.
• Knijn, T., Kremer, M. (1997). Gender and the caring dimension of welfare states: Towards inclusive citizenship. Social Politics, 4,328-361.[Abstract/Free Full Text]
• Lister, R. (1997). Citizenship: Feminist perspectives. Washington Square, NY: NYU Press.
• Loh, E. S. (1986). Productivity differences and the marriage wage premium for white males. Journal of Human Resources, 31,566-589.
• Lundberg, S., Rose, E. (2000). Parenthood and the earnings of married men and women. Labor Economics, 7,689-710.
• Manning, W., Newhouse, J., Duan, N., Leibowitz, A. (1987). Health insurance and randomized experiment. The American Economic Review, 77,251-277.[Medline]
• Mitchell, O., Hammond, P., Rappaport, A. (2000). Forecasting retirement needs. Philadelphia: Pension Research Council.
• Orloff, A. S. (1993). Gender and the social rights of citizenship. American Sociological Review, 58,303-328.
• Porter, K., Larin, K., Primus, A. (1999). Social Security and poverty among the elderly. Washington, DC: Center on Budget and Policy Priorities.
• Rubin, D. (1987). Multiple imputation for nonresponse in surveys. New York: John Wiley.
• Secombe, W. (1974). The housewife and her labour under capitalism. New Left Review, 83,(Jan.–Feb.), 3-24.
• Smeeding, T., Estes, C., Glasse, L. (1999). Social Security reform and older women: Improving the system (Paper No. 22). Syracuse, NY: Syracuse University.
• Smith, J., Ward, M. (1985). Time series growth in women's labor force participation. Journal of Labor Economics, 39,59-90.
• Thompson, L., Carasso, A. (2002). Social Security and the treatment of families: How does the U.S. compare to other developed countries. In M. Favreault, F. Summartino, & C. Eugenesteurle (Eds.), Social Security and the family (pp. 123–176). Washington, DC: Urban Institute Press.
• Waldfogel, J. (1997). Understanding the family gap in pay for women with children. Journal of Economic Perspectives, 12,137-156.

This article has been cited by other articles:

				J. L. Angel, M. A. Jimenez, and R. J. Angel The Economic Consequences of Widowhood for Older Minority Women Gerontologist, April 1, 2007; 47(2): 224 - 234. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation