Women and Economic Security in Retirement: Implications for Social Security Reform Janet Spratlin Karen C. Holden University of Wisconsin–Madison

ABSTRACT: Gender-related parenting roles lower incomes of women during the working-age years and in retirement years. National retirement income programs contribute to this gender-related difference in retirement income to the degree that they consider lifetime earnings in calculating benefits. This article examines two national pension systems: those of Australia and Sweden. The pension systems of these two countries often are considered as the extremes of pension generosity, and they are based upon different philosophies about the role of government in insuring against economic risk. Due to fiscal issues and the changing roles of women in the labor market, both have reformed their systems. How system reforms can balance the role of the government in insuring against income risk and the growing financial independence of women in the work force are discussed. KEY WORDS:

national pensions, Australia, Sweden, social security reform, women.

Introduction This article discusses the gender-related income risks that are associated with marital and parenting roles and asks how the benefits of public retirement income programs perpetuate or modify those genJanet Spratlin is research fellow in the Consumer Science Department at the University of Wisconsin. Her research interests include care giving responsibilities assumed during women’s mid-life working years, how these responsibilities impact the long-term financial security of women, and how personal decisions and public policy can reinforce or counteract the impact of these responsibilities on the financial security of women. She may be reached by e-mail at: spratlin6facstaff.wisc.edu. Karen C. Holden is professor of Public Affairs and Consumer Science at the University of Wisconsin-Madison. Her research interests include social insurance specifically regarding the welfare of the aged and the effect of social security and pension policy on economic status after retirement and widowhood. Other research interests include disability, welfare reform, mandatory retirement policies, and risk of nursing home care. She may be reached at 1225 Observatory Dr., Madison, WI 53706. Journal of Family and Economic Issues, Vol. 21(1), Spring 2000 q 2000 Human Sciences Press, Inc.

37

38

Journal of Family and Economic Issues

der-related economic risks into retirement. The objective of this discussion is to explore whether and how different benefit constructs, based upon different philosophies about appropriate roles of public and private insurance, lead to different outcomes for single elderly women living alone. First, the sources of gender-related benefit differences in old age are described. Then, the features of social insurance systems that reflect or moderate gender-related income differences that otherwise would exist due to pre-retirement earnings differences due to time spent out of the workforce are discussed. A brief discussion of the demographic and economic changes that have motivated public pension reform and how women’s changing work roles have been used to justify changes in coverage by and benefits paid to women as widows follows. The primary focus of this article is whether systems that reflect divergent national precepts about the role of government in modifying private income risks result in different income security outcomes for older persons who spent time out of the work force. Different income outcomes are a function of different assumptions about government responsibility, which lead to programmatic approaches that cover work reduction patterns differently. Alternatively, formal program rules may represent different ways of providing comparable insurance against the same earnings contingencies. Comparable coverage can occur if different programmatic approaches, while paying benefits in different ways and from different sources, cover precisely the same risks at the same level. A description of two national pension systems follows. These two pension systems were selected because they present contrasting philosophies about the appropriate role of government in protecting women as retirees and widows against the income consequences of care-giving earlier in their lives. This article concludes with a discussion of the ways in which systems, as they move from the traditional protections provided to women as economically dependent individuals to mixed systems that include individualized accounts, can provide protection for those women and men who reduce labor force activities in order to raise children.1 This discussion includes reference to the debate over Social Security reform in the United States. In evaluating how women fare under social retirement systems, this article focuses on the following: 1. The extent to which the system provides a benefit sufficient to raise beneficiaries above some minimum level of income;

Janet Spratlin and Karen C. Holden

39

2. The extent to which the system insulates retirement benefits from the impact of earlier care-giving on earnings; and 3. The extent to which the system protects women against the consequences of their longer life expectancies, including the effects on annuity values, of widowhood, and of inflation. Public Retirement Programs and the Economic Status of Elderly Women: An Overview A key indicator of the effectiveness of an income maintenance program is its ability to reduce the risk of declining into poverty. Using this measure, public retirement programs are among the most successful income maintenance programs that exist. For example, over 35% of all U.S. residents age 65 and older are estimated to have been living below the official poverty threshold in 1969. By 1996, that percentage had fallen to just over 9% (Social Security Administration, 1997; Grad, 1996). Younger American households, especially those with children, now are more likely than the elderly to be poor.2 The same pattern of elderly households doing better than households with heads under age 30 is also generally true across European nations (Smeeding & Sullivan, 1998). However, widespread gains in economic well-being among the elderly are not found equally among men and women. In most countries, elderly women living alone are more likely than the average pensioner to be poor (Holden & Smock, 1991; Smeeding, 1997; Shaver, 1996). Table 1 compares the poverty rates for seven countries, with poverty defined as family income being less than 50% of a nation’s median disposable income (Smeeding, 1997). TABLE 1 Percent of the Elderly Population in Poverty Country

US

UK

All elderly 22.7% 30.5% Elderly women living alone 43.2% 50.1% Year of data 1994 1991

Australia Canada Germany Netherlands Sweden 28.6%

7.1%

8.1%

4.4%

8.4%

62.1% 1989

16.2% 1991

12.7% 1989

3.4% 1991

14.7% 1992

Note: Adapted from “Reshuffling Responsibilities in Old Age: The United States in a Comparative Perspective,” by Smeeding, T., 1997, LIS Working Paper No. 153, Available at: http://lissp.ceps.lu/wpapersd.htm.

40

Journal of Family and Economic Issues

These data reveal considerable cross-country variation in the difference between the poverty rate for the elderly in general and the poverty rate for elderly women living alone. Australia has one of the largest differences, with a poverty rate among single elderly women of 62% compared to 29% for all elderly. At the other end of the spectrum is the Netherlands where elderly women living alone are less likely than the average pensioner to be poor. In terms of these two poverty measures (i.e., the level of poverty among the elderly and the difference between the percent of retired women living alone and the percent of retirees as a whole who are poor), Sweden is well above the Netherlands and is more like other Western European countries and Canada.3 The reason for these cross-national differences is not well understood. While variation in the type and generosity of national old-age income programs is assumed to be the major factor (Shaver, 1996; Smeeding, 1997), private insurance decisions that increase economic risk also are important (Holden & Nicholson, 1998). Private and public pension systems determine a woman’s well-being in retirement as a result of how much she has worked over her work life.

The Keys to Older Women’s Financial Vulnerability: Family Roles and Survivorship Why do older women living alone often end up in an economically worse position than either men living alone or married couples? The answer to this question is linked to two critical gender differences: (a) women’s different family roles and the impact these roles have on women’s work patterns, earnings, and their ability to accumulate retirement resources; and (b) women’s longer life-expectancies, which means that they face a longer retirement period over which resources must be stretched and that they face a greater likelihood than married men of income loss due to a spouse’s death. Different family roles reflect how husbands and wives share childrearing and earnings responsibilities. Traditionally, women primarily have been responsible for child-rearing, which leads to lower women’s labor force participation rates and greater part-time work during the child-rearing years. For example, in the U.S. 68% of married women with at least one child under the age of 18 were employed in 1998, 29% of them part-time. In comparison, 93% of married men with children under 18 were employed, and only 3% were part-time workers

Janet Spratlin and Karen C. Holden

41

(Bureau of Labor Statistics, 1999). While married couples can share the responsibilities of income generation and child-care, the income consequences of child-rearing will be more fully borne by unmarried mothers.4 This lesser attachment to the labor market by mothers interacts with women’s lower average market wages to limit their lifetime earnings, even after they return to work as full-time, full-year workers. The extent to which gender differences in lifetime earnings then are translated into gender differences in retirement incomes depends upon how closely public and private retirement income is tied to prior labor force attachment and earnings history. In strictly earningsbased private or public pension systems, workers with fewer years of covered work and lower wages receive lower retirement benefits. Retirement systems can compensate for the retirement-benefit effects of time out of the work force by shortening the earnings period used to calculate benefits, by providing more generous replacement rates for low income earners, or by guaranteeing minimum benefit levels.5 Longer life expectancies. In 1996, a 65-year-old woman in the U.S. could expect to live on average another 19 years compared to 15.7 years for a male the same age (Anderson, 1998). The longer expectation of life among women means that accumulated retirement resources must be spread over more years of retirement. Thus, when retirement annuities are calculated using sex-based actuarial tables, women will receive lower monthly benefits than will men the same age with identical account totals. This is likely to be the case for annuities purchased with personal savings or for annuities purchased outside a pension plan with lump-sum distributions.6 When lifetime annuities are purchased, women are disadvantaged further because private annuity markets are priced using sex-specific, risk-based factors. Many countries mandate annuitization of retirement accounts because without such regulation annuity prices tend to be driven by adverse selection.7 In contrast, defined-benefit plans calculate benefits using formulas that usually include covered earnings and years of service, and they typically pay equal benefits to women and men with identical covered work histories. Similarly, actuarial-based calculations will lead to lower benefits paid during the lifetime of the primary beneficiary when a survivor benefit is selected, whether that choice is made within the context of a defined-benefit or defined contribution plan.8 In contrast, most government pension schemes pay survivor benefits, usually as a fixed percentage of the worker’s benefit, without any adjustment in the

42

Journal of Family and Economic Issues

amount paid to the retired worker to offset the longer period over which joint annuities must be paid. Thus, one of the key advantages to women, as covered workers or widows of covered workers under government-sponsored defined-benefit retirement systems, is that the individual mortality and survivorship risk is spread over the entire population, which, on average, benefits women. Finally, over the longer lifetimes of women, inflation will exact a larger toll on the real value of pension income. After retirement, the pension benefits will decline in purchasing power if not adjusted fully for price increases and will fall relative to the incomes of the working population if not adjusted for productivity gains. In most government pension programs, there are adjustments to offset inflation, but pensioners are less likely to share through wage-based adjustments in the productivity gains enjoyed by the working-age population.

Forces of Reform in Public Retirement Programs Most national pension systems have been funded on a pay-as-yougo (PAYG) basis, which taxes current workers and employees either directly through payroll taxes or indirectly through the general tax system to pay pensions of current pensioners. Such systems can support increases in benefits to retirees without raising tax rates on the working-age population as long as the tax base expands. During the 1950s and 1960s, economic conditions and initially low social security tax rates in most of the developed world supported expansion of benefits. By the 1970s, these happy conditions were coming to an end. Economic growth had stagnated, and workers resisted increases in what was no longer a seemingly low tax rate. The retiree population expanded more rapidly than the working population, which was a demographic shift driven by lower fertility, longer life spans, and earlier retirement ages in most of the developed world (Lee & Tuljapurkar, 1997; Weaver, 1998; Wise, 1997). The consequent pressure to raise taxes to support a growing population of retirees led to national concerns about the consequence of increasing payroll taxes for national savings, economic growth, and international competitiveness (Gramlich, 1997). The same demographic and economic forces that threaten the financial viability of national pension programs also have changed the distribution of pension costs and benefits across different types of households. Most wives, even those who are mothers of small chil-

Janet Spratlin and Karen C. Holden

43

dren, now are in the work force. These women are far less likely to be solely dependent upon their husbands for old-age support than were their own mothers. Nevertheless, survivor pensions continue to be a significant protection against poverty among today’s elderly women (Gramlich, 1998). At the same time, a significant proportion of women are raising children as never married or divorced parents. For many of these women, family responsibilities will continue to affect their earnings adversely and reduce the chances of sharing in the retirement incomes of a spouse. These contrasting economic forces, one strengthening and the other weakening the retirement security of women, bring into question whether and how pension reform should preserve redistributional components designed primarily to benefit women who are largely dependent upon their husbands for financial security.

The Traditional Solutions to Wives’ Dependency: Survivor Benefits and Universal Pensions The fundamental value judgment implicit in any government-provided old-age retirement system is that society has an obligation to assure income to its citizens. Governments may do this through private market regulation to insure that promises made by private employers and insurers are fulfilled. Regulation alone would leave individuals subject to private market risks, including the consequences of their own (or their spouses’) choices to remain uninsured. As a consequence, most nations provide some form of public pension to insure against retirement-income risks and errors in individual judgment about the size of that risk (Gramlich, 1997). Public pension systems can be classified into two broad groups: (a) earnings-related systems that directly link benefits to prior earnings; and (b) those that guarantee each individual a minimum benefit regardless of that individual’s earnings history. Among earnings-related retirement systems, the most frequently adopted solution to the oldage income risks associated with earlier labor force interruptions has been to provide benefits to one spouse based upon the earnings record of the spouse with a more extensive work history.9 Such a system allows women as widows and sometimes wives to share in the retirement benefits of their husbands regardless of their own earnings histories. Women can expect to share in this way in the (perhaps then market-specialized, higher) earnings-based retirement benefits of

44

Journal of Family and Economic Issues

their husbands after their husbands’ deaths as they did before. To this end, most earnings-related systems have incorporated mandated survivor benefits and, in some cases, spouse benefits, and many also have some form of mandated benefit-sharing among divorced spouses (Social Security Administration, 1999).10 The second system type offsets the differential effect on old-age economic status of earlier interruptions in market earnings by paying either a universal demogrant, that is paid without regard to income or other means, or through a benefit that is means-tested, which is targeted to those individuals and families considered most in need of additional income. Both demogrant and means-tested systems insure a minimum living standard without specifically differentiating or targeting patterns of earnings or reasons (such as child-rearing) for interrupted or low earnings. Likewise, setting a minimum per-person benefit would treat survivors and never-married persons alike without specifically addressing patterns of benefit bequesting. In a universal or means-tested grant system, all citizens are assured a minimum level of income, regardless of wage earnings. However, smoothing inequalities in earnings and retirement benefits beyond that minimum (due in part, to different labor participation choices and earnings options) are deemed not to be a government issue. Individuals with higher earnings bear responsibility for insuring against the loss of that income upon retirement; a labor force decision that increases the chances of receiving only the guarantee is a personal choice. In contrast, in earnings-based systems, earnings are insured (though typically at lower rates as earnings rise, often up to a maximum) against the income consequences of their loss. Systems with survivor benefits link the income guarantee to the earnings of deceased spouses, with the result that widows (and in some countries widowers) of higher earners receive higher survivor benefits. As nations explore ways to reduce the social costs of government retirement programs, critical questions arise:

• How can reformed systems insure against the greater retirement income risks facing women?

• Is it possible to design earnings-based retirement systems that •

reduce the risk of later-age poverty for women whose earnings were interrupted due to child-rearing? If traditional survivor pensions no longer fit the needs of a society in which most women work outside the home, what, if anything, should replace them?

Janet Spratlin and Karen C. Holden

45

The next section explores how women are likely to fare in their later years under two different insurance schemes that represent two different approaches to income security: the national pension systems of Australia and Sweden. These two countries were selected because both recently have reformed their public retirement systems in ways that significantly increase their reliance on defined contribution programs, in which benefits received are linked closely to earningsrelated contributions and investment returns over the course of an individual’s working life. Despite this broad reform similarity, the particular reforms reflect: (a) each nations’ contrasting long-term philosophies about the appropriate role of government in insuring against retirement income declines; and (b) the variety of income maintenance structures already in place. The question, to which the remainder of the article is devoted, is how these different approaches to pension reform are likely to affect income security in old age for women who spend some years out of the work force. For each system, the process of determining benefits is described. The extent to which the system and its separate tiers protect women against retirement income risks associated with longer life expectancy and with time out of the workforce for childcare also are described. Tables 2 and 3 summarize the main features of the current reformed systems in these two countries.

Australia: Supplementing a Means-Tested Minimum Pension with Mandated Private Coverage The System Australia provides a minimum guarantee to all citizens through its means-tested Age Pension, which is funded by general revenue. Prior to 1992, this was Australia’s only universal coverage pension program at the national level. The year 1992 also was the year in which the Superannuation Guarantee was implemented. The Superannuation Guarantee mandated that all employers make contributions on behalf of their employees to the Superannuation funds of the employer’s choice.11 Contributions (currently 7% of pre-tax earnings, rising to 9% by 2002) are placed in individual accounts and invested on behalf of the employees. All workers earning above a monthly minimum (A$450 in 1999) must participate, excluding the self-employed. The Age Pension, which was not changed as part of the reform, will

46

Journal of Family and Economic Issues

TABLE 2 Minimum Guarantee Tier: Australia and Sweden Australia: Age pension Long-standing program that was not modified by recent reforms Funded out of general revenue Means-tested against other pension and non-pension income Means-tested against assets (not including owner-occupied housing) Income and asset limits are indexed to annual movements in the CPI Maximum age pension set at 25% of male avg. earnings f/ individual 40% of male average earnings f/ a couple Maximum amounts indexed to annual movements in Male Average Earnings.

Sweden: Guaranteed minimum pension Replaces the national basic pension that was provided to all who lived in Sweden 40 years or worked there 30 years Funded out of payroll tax earmarked for this purpose and separate from the two-tiered earnings-related pension Cleared (means-tested) against other government pension income No asset limitations N/A Guaranteed minimum pension set at: 2.13* (Price base amount) approximately US$9,700 3.8*(Price base amount) approximately US$17,300 Guaranteed minimum indexed to annual movements in the CPI

continue to provide the basic guarantee for all eligible seniors, set at 25% of male average earnings for a single retiree and 40% of male average earnings for a retired couple. Means-testing reduces benefits when the recipient has other income, including from Superannuation plans and assets, excluding housing. Pension Levels and Poverty Risk The Age Pension currently provides at least some income support to over 80% of Australia’s elderly population (i.e., those age 65 or older), with 68% receiving the full benefit amount. Not surprisingly, women currently are more likely than elderly couples to qualify for Australia’s Age Pension, with 91% of women and 75% of elderly couples receiving benefits. These benefits account for a larger fraction of women’s total income, with benefits accounting for 77% of total income of women and 50% of total income of elderly couples (Shaver, 1996). This difference in receipt and income composition is a reflection of women pensioners’ more intermittent work histories and their lower likelihood of having been employed in jobs that were covered by private sector Superannuation plans.

Janet Spratlin and Karen C. Holden

47

TABLE 3 Defined-Contribution Tiers: Australia and Sweden Australia: Superannuation Guarantee Government regulated, privately managed defined contribution pensions. Newly implemented program patterned on voluntary occupational Superannuation plans. Fully funded by “employer contributions” currently set at 7% of wages, scheduled to rise to 9% in 2002. Contributions made to privately managed Superannuation Funds chosen by the employer. Deposited into individual, private accounts in the worker’s name. Growth in account balances are directly related to the investment performance of the applicable Superannuation Funds. Funds available at time of retirement are wholly dependent on contributions and investment returns on those contributions.

Assets can be withdrawn as a lump sum at age 55.

Sweden: PAYG tier

Sweden: Reserve accounts

Defined contribution, PAYG government pension.

Premium reserve: Government managed, privatized accounts.

Replaced defined-benefit pension in which benefits were based on contributions for best 15 years. PAYG funding from payroll tax of 16% of wages.

Newly implemented program.

Taxes on current workers go to pay the pensions of current retirees.

Contributions made to funds specified by the individual workers.

Record-keeping system accounts for individual’s contributions.

Deposited into individual, government regulated, privately managed accounts in the worker’s name. Growth in account balances is directly related to the investment performance of the individually selected funds.

Accounts are credited by the government with an annual interest reflecting the economic growth of the Swedish economy. Pensions provided to individual retirees are dependent on: a) Their contributions, b) The interest “earned” each year on the “account balances,” and c) Inherited gains (contributions of workers who die before they retire are distributed among all workers born in the same year). Pension is paid out only as an annuity.

Fully funded from payroll tax of 2.5% of wages.

Funds available at time of retirement are wholly dependent on contributions and investment returns on those contributions.

If taken out as income, funds must be annuitized.

48

Journal of Family and Economic Issues

TABLE 3 (Continued ) Australia: Superannuation Guarantee

Sweden: PAYG tier

No requirement that these funds be annuitized. Voluntary annuity market subject to adverse selection. Annuities are calculated using single sex actuarial tables. No mandated survivor benefits, Few couples purchase joint and survivor annuities in voluntary annuity market. No mandated sharing of assets in the event of divorce.

Mandatory annuitization eliminates adverse selection.

Distribution as a lump sum allows: —Spouses to plan consumption according to own evaluation of morbidity, mortality risks. —Spouses to bequeath pension assets to one another. —Retirees to spend down assets before death.

Individuals can choose to withdraw only part of annuitized pension during early years leaving a larger pension for later. Flexibility allows couples to draw down respective pensions at rates that reflect own respective life expectancies.

Annuities are calculated using unisex actuarial tables. No survivor benefits allowed. No mandated sharing of assets in the event of divorce.

Sweden: Reserve accounts

Annuities are calculated using unisex actuarial tables. • Prior to retirement can use funds to buy survivors’ insurance. • At time of retirement can use funds to purchase a joint & survivor annuity. • Can also bequeath funds to spouse or children, who must withdraw them as an annuity. Tier II pension rights can be transferred to a spouse or registered partner as these rights accrue (but not retroactively.)

The degree of success of the Age Pension in lifting Australian pensioners out of poverty depends upon the poverty threshold. When the poverty level is set at 40% of median income, only 2.9% of elderly couples and 3.5% of elderly Australian women living alone are poor. The Age Pension uniformly raises incomes to a level that closely corresponds with this threshold. These levels of poverty compare favorably with other European countries and are below poverty rates in the U.K. and Germany. When, however, the poverty threshold used to judge relative economic well-being is raised, the increase in the proportion of Australian beneficiaries falling below these higher thresholds is much greater than that in other nations, as is the disparity between poorer single women and elderly couples (Shaver, 1996).12

Janet Spratlin and Karen C. Holden

49

Survivorship and Annuitization The Age Pension and the Superannuation Guarantee do not include provisions for survivor benefits. The absence of a survivorship provision in the Age Pension is consistent with its minimum income guarantee to all citizens regardless of earnings history or taxes paid.13 In contrast, the Superannuation Guarantee is related directly to an individual’s lifetime earnings, which means that benefits will be lower for parents who adjust labor force activities for child-rearing. Yet the Superannuation Guarantee program operates like an individual forced savings program, making no requirement that savings be taken out as an annuity or that, if annuitized, some provision be made by married workers for the surviving spouse.14 Means-testing in the widely received Age Pension is an incentive for Superannuation beneficiaries to assure their likely eligibility for the Age Pension by taking their Superannuation as a lump sum and spending down or investing in assets (such as owner-occupied housing) that are not means-tested against the Age Pension. Even in the absence of mandated survivor benefits, Australian women are expected to benefit significantly from the introduction of the Superannuation Guarantee. Women who spend some time in the work force now will have universal pension coverage and can expect to enjoy retirement incomes (often significantly) above those provided by the Age Pension. This is consistent with the data reported in Bateman and Piggott (1998) showing that the introduction of the mandatory Superannuation program has resulted in a much larger percentage increase in Superannuation funds among women than among men. At the same time, however, these retirement savings will be related strictly to earnings and investment. The Superannuation Guarantee provides no offset for the lack of Superannuation contributions during years spent out of the work force or for the reduction in Superannuation contributions due to the negative impact of interruptions in paid work on subsequent wages. Consequently, even though Australian women entering the work force today are likely to have higher retirement incomes than their mothers, as long as women continue to be more likely than men to adjust their work lives to accommodate family responsibilities, women’s earnings before retirement and their incomes during retirement are likely to continue to trail those of men. Not requiring funds to be annuitized provides both flexibility and risk. It enables married men and women to bequeath to their respec-

50

Journal of Family and Economic Issues

tive spouses whatever remains of their own retirement assets at the time of their death or to choose a consumption plan most suited to their individual morbidity and mortality risk. It also, however, allows couples to spend down their combined assets during the period when both are alive, leaving the surviving spouse, who is usually the wife, solely dependent upon the Age Pension during widowhood. For this reason, Bateman and Piggott (1998) argue in favor of establishing shared Superannuation Guarantee property rights and requiring that each spouse select joint and survivor annuities. Despite the government’s introduction of a variety of incentives designed to encourage retirees to annuitize their income, only about 10% of the value of Superannuation benefits has been paid in this form (Doyle & Piggott, 1998). This suggests that, indeed, the public perceives annuitization in the private market as too expensive for the value received. Longer Life and Inflation Australia’s Age Pension is indexed to the CPI, providing inflation protection to the share of retiree’s income that derives from this source. Linking its maximum benefit to average male wages also allows some limited sharing by elderly pensioners in productivity gains. The Superannuation funds will grow with investment earnings. Thus, to the extent that earnings in these funds reflect economy-wide price and productivity gains, retirees are protected against real declines in relative retirement benefits. The particular form in which the fund is withdrawn will determine what happens after retirement. The Australian Annuity market offers price indexed annuities, and the means-testing provisions of the Age Pension encourage their purchase,15 but, as was discussed earlier, the adverse selection present in voluntary annuity markets and gender-based annuity calculations reduce the attractiveness of this option.

Sweden: An Explicit Focus on the Cost of Children The System The Income-Related Old Age Pensions Act, which was adopted in 1998, represents a conscious shift from a long-standing compulsory, government-managed, nationally uniform, PAYG defined-benefit pension plan to a compulsory, two-tiered, defined-contribution system.

Janet Spratlin and Karen C. Holden

51

The larger tier will continue to be a government-run, PAYG system but now is converted to a defined-contribution system, with a second tier that is a pre-funded, privately managed, individual retirement account program.16 The bulk of the earnings tax (16% out of the 18.5%) will be used to finance the PAYG tier in which workers’ contributions in any given year are used to pay the pensions of current retirees. An accounting will be kept for each individual of the contributions made to the system as entitlements on the individual’s pension account. This account is backed by a claim on present and future contributors to the PAYG system. Workers’ contributions into the system grant each of them a claim equal to: 1. The amount they contributed over their lifetime, plus 2. An interest return that reflects the growth or decline of the Swedish economy over the period during which the insured were in the workforce, plus 3. Inherited gains, which are the contributions from people who die each year before reaching retirement age being distributed to insured persons of the same age. The remaining 2.5% of the earnings tax will be deposited into the worker’s own Premium Reserve account. This is a pre-funded, privatized account system in which the insured will be able to choose an investment manager for this part of their pension. The pension amount received from each premium reserve account will depend upon the actual returns achieved by the particular manager chosen to invest the funds. There is no government-guaranteed minimum return on funds placed in these individual accounts. Pension Rights for Child Care As part of this reform, Sweden has implemented an innovative adjustment to earnings records, specifically designed to reduce the impact of parenting on the care-giving parent’s pension income. Creating “pension rights for childcare,” the state contributes to the appropriate parent’s pension account 18.5% of the appropriate “pensionable amount,” defined as the highest of three different calculations: (a) the amount needed to raise the qualifying parent’s actual pensionable income during childcare years to their pensionable income for the year prior to the year during which the child is born, (b)

52

Journal of Family and Economic Issues

the amount needed to raise the qualifying parent’s actual pensionable income during childcare years to 75% of the average income of all insured persons under the age of 65, and (c) an amount equal to one income base amount.17 Contributions based upon the pensionable amounts are funded by the social insurance system that is responsible for supporting families with children.18 Parents (only one of which is eligible) will be eligible for these contributions during their child’s first four years. Parents who have more than one child during this four-year period will be eligible for pension rights for only one of the children at a time, although parents of two or more children, who space their children with this in mind, can extend their eligibility significantly beyond four years. Notably, parents do not have to stay home or work part-time to be eligible for childcare-related pension rights. Even parents who work full-time, with earnings equal to or greater than 75% of the average income of persons under 65, will be eligible for pension rights equal to one base amount to compensate for the extra work involved in parenthood. Pension Level and Poverty Risk Sweden’s pre-reform defined-benefit system provided a universal, basic benefit plus an earnings-related pension, both of which were funded by a payroll tax. The reformed system continues to provide a minimum guarantee but offsets this amount against other government-related pension income. In 1998, the minimum guaranteed pension was set at 2.13 times the current price base amount or approximately $US 9,700 for an individual and 3.8 times the current price base amount or approximately $US 17,300 for a married couple.19 Recipients of the full Minimum Pension typically also would be eligible for a housing allowance, and the combined amount would be taxable income. Virtually all comparative studies show that Sweden’s poverty rates are among the lowest among the developed world regardless of the poverty threshold or equivalency scales used (Kenworthy, 1998; Shaver, 1996). Receipt of social transfer payments is universal among the elderly, and these payments account for virtually all of the income of the elderly (Shaver, 1996). Nevertheless, single women still are more likely to qualify for means-tested benefits than are couples (i.e., 49% of women versus 13% of couples qualify), and they are more likely to be poor than are couples of the same age (see Table 1).

Janet Spratlin and Karen C. Holden

53

Survivorship and Annuitization After extensive debate, the architects of the reformed Swedish pension system eliminated mandated survivorship pensions in the PAYG, earnings-related tier. Under the new system, husbands and wives each will draw an annuity based upon their own credited accumulations, and, following the death of a spouse, survivors will be expected to use of their own annuity to pay their own living expenses, which, however, may have been augmented by implicit wage credits for child care that had been credited to the individual’s account. The implicit value judgment is that, except for a limited period when first caring for children, both men and women should be in the labor force working and contributing to their own retirement savings. Even though the PAYG tier does not pay survivor pensions, annuitization is required. Couples do, however, have the ability to adjust to the rate of payout of the annuity in ways that increase protection to surviving spouses or registered partners. A worker may vary the rate of withdrawal over time, for example, drawing down only part of the pension early on and letting the rest accumulate until later. While the intent of this flexibility was to encourage retirement-age workers to delay retirement or to work part-time, it also can be used by a couple to vary pension withdrawal in a way that reflects their own estimates of their respective life expectancies and the consumption needs of the probable survivor. For example, if the husband expects to die first, the couple could choose to annuitize his pension but annuitize none or only a fraction of her pension during his expected lifetime. This would allow for the wife’s pension, when she finally decides to resume a 100% annuity, to be larger than it would have been had she drawn down 100% from the very beginning. As in Australia, this flexibility allows couples to adapt income streams to expected retirement income needs. Unlike Australia, required annuitization protects workers and their survivors from the consequences of spending down retirement assets too quickly. Likewise, in the pre-funded portion of the program, there are no subsidized or mandated survivor benefits, but couples are able to share the benefits from their Premium Reserve pensions in several ways: 1. They are entitled to transfer their pension rights to a spouse or registered partner as these pension rights accrue; 2. Upon retirement, the beneficiary can opt for a traditional single-

54

Journal of Family and Economic Issues

life annuity that will pay a fixed monthly amount for the duration of his or her life (computed on a unisex life table), or for an actuarially equivalent joint and survivor annuity that would pay a smaller benefit while the primary beneficiary is still alive; 3. The insured person can designate the pre-funded pension as a survivor’s benefit, in which case the account would remain invested until the death of the insured, at which point it would be transferred to the insured’s spouse or child, who then would be required to withdraw the funds as an annuity; and 4. Pre-retirement participants with children under 20 or a current or former spouse can use the assets in the pre-funded account to purchase five-year term life insurance. Life Expectancy Effects As noted above, Swedish women benefit from the requirement that the government-run PAYG first tier pension be withdrawn as an annuity that is calculated using unisex actuarial tables. Similarly, even though there are various withdrawal options for the tier two, prefunded accounts, if these assets are withdrawn as income (rather than used to buy insurance for survivors, etc.), they also must be annuitized using unisex actuarial tables. Sweden’s PAYG first tier pension also provides pensionners with excellent protection against inflation. Not only are beneficiaries protected against loss in the real value of their pensions due to price increases, but they also can expect to share at least partially in the nation’s economic growth. Sweden assumed an annual increase in economic growth of 1.6% in calculating the monthly benefit at the time of retirement. Annual adjustments to the monthly benefit check reflect the extent to which the previous year’s real economic growth deviated from the assumed 1.6%. The economic adjustment index will have the effect of increasing pensions by the rate of inflation plus or minus the number of percentage units that the growth of real wages deviates from the 1.6% norm. Under this process, pensioners will receive full compensation for the increase in inflation as long as real wages grow as fast as inflation. If real wages do not keep up with the growth in prices, the adjustment in pensioners’ benefits will be limited to the growth in real wages. If, alternatively, real wages grow faster than inflation, pensioners will receive compensation for full inflation, and they also will share some of the growth in real wages enjoyed by the economically active.

Janet Spratlin and Karen C. Holden

55

Discussion: The Lessons for Reform Despite enormous social and demographic changes and the growth of family-friendly policies that ease parent’s participation in work force, under virtually all social security schemes the parent who assumes the bulk of child-rearing and other care-giving responsibilities will continue to see these non-market activities reflected in his or her subsequent lower earnings and lower earnings-related pensions. Nevertheless, women’s increasing economic independence makes dependency-based benefits paid to wives and widows seem anachronistic, if not inequitable and unfair. It is important that discussions of social security reform balance a nation’s interest in maintaining equity across beneficiaries and not jeopardize the financial security of parents who chose and may have been encouraged by the larger society to devote time to child-rearing. As social security reform options are debated in the U.S. and elsewhere, it is useful to examine the approaches taken by other countries in balancing the greater financial independence of women against the fact that child-rearing continues to reduce labor force participation and earnings of parents. Prior to the early 1990s, Australia and Sweden had public retirement systems that reflected radically different philosophies about the responsibility of government regarding the financial well-being of elderly citizens. Australia guaranteed a minimum benefit to all elderly and left any remaining insurance choices to the individual. Sweden also guaranteed a minimum, but in addition, workers and their survivors were covered by a relatively generous public defined-benefit pension. The absence of a universal, public earnings-related retirement program in Australia typically is named as the source of higher poverty rates among Australia’s elderly women living alone. The broad features of the reformed systems in Sweden and Australia systems are much alike. Both countries have maintained a minimum guarantee, but both now have a second defined-contribution tier that has no required survivor benefits but does allow considerable flexibility to pensioners in how their pension is paid. Given this broad similarity, does it make a difference whether that second tier is provided by the public or an unregulated private sector? Are the other differences in the system important to the well-being of women, especially to those who spend some time out the workforce? In the final part of this article, conclusions are drawn about program features, which likely will be important to the security of older women in these

56

Journal of Family and Economic Issues

two countries. This discussion is organized around the three questions: 1. Success in raising pensioners above a minimum income level. With the addition of its mandatory Superannuation Program, Australia’s retirement system has moved closer to the social insurance systems in northern Europe and Canada, most of which combine a minimum pension with an earnings-related pension. Unlike the U.S. Social Security program, which combines redistributive and insurance features in its single-tiered, earnings-related Social Security program, a dual scheme can separate these functions more cleanly. The minimum benefit first tier has facilitated the adoption of strict earnings-related second tier benefits by both countries, by providing a safety net that protects non-workers and low income workers and maintaining the important function of assuring a given level of economic well-being to all citizens. In general, the success of these dual systems in lifting elderly residents out of poverty depends most directly upon the universality and generosity of the first tier. Both Sweden (through its Minimum Pension) and Australia (through its Age Pension) provide a universal minimum benefit that, using a poverty threshold equal to 40% of median income, results in low levels of poverty among elderly women, which are comparable to those in other advanced nations. However, because the Swedish minimum is set higher relative to population income medians than is the minimum in Australia’s Age Pension, and because this benefit is offset against other public transfers only, low poverty rates are maintained in Sweden even when poverty thresholds are raised. In contrast, the use of higher thresholds results in much higher poverty rates in Australia. 2. The extent to which the system insulates pension benefits from the impact of care giving. Minimum benefits insulate against earnings loss only at low income levels. Above this guarantee, an earningsrelated tier may provide a constant replacement rate or provide higher replacement for lower earners. By shifting to defined contribution plans, nations eschew the ability to retain redistribution as a component of that tier. As is true in Australia and in Sweden, accumulations in the defined-contribution tier will be smaller when a worker spends time out of the workforce. Whereas in the pre-reform Swedish system, benefits were calculated using a worker’s 15 years of highest earnings, benefits now will be based upon contribution ac-

Janet Spratlin and Karen C. Holden

57

counts and attributed earnings, in effect extending pension coverage to all earning years. Although this change penalizes parents who reduce work to raise children, it was viewed as fairer to long-term workers who in the pre-reform system often paid more taxes and received fewer benefits than workers whose careers were compressed into shorter time spans. It is likely that the introduction of the Superannuation Guarantee in Australia will increase the incomes of Australian retirees and reduce but not eliminate the gender imbalance among retirees dependent upon the Age Pension. Australia’s mandated pension coverage provides coverage to all workers, including those women, who were in occupations not traditionally covered by pensions. Nevertheless, as long as women continue to be more likely than men to reduce their labor force activity or choose lower paying jobs that better accommodate family life, women will continue to fall lower in the earnings distribution and, therefore, lower in the retirement income distribution. While redistribution may be a less prominent component of defined-contribution plans, Swedish reforms have shown that compensation for time out of the workforce can be incorporated into the system. The Pension Credits for Children make a major contribution toward reducing the gender gap in benefits when those benefits are based upon prior earnings. Through this program, the Swedish government explicitly acknowledges the pension costs to parents who reduce their work hours to care for children. Even though these credits are provided for only a few years and the credits may not fully compensate for lost earnings, they do make a substantial contribution to the pensions of parents, most of whom are women. Because these credits are granted to relatively young women bearing children, the compounding over 30 or more years of contributions of 18.5% of an amount equal to 75% of the average income of workers under 65 even if paid for only four years, can represent a significant contribution to the income of retired persons. 3. The extent to which the system protects women against the consequences of their longer life expectancies. Annuitization risks. Sweden has mandated that the PAYG pensions and all annuities bought with premium reserve be annuitized using unisex actuarial factors. In contrast, Australia does not mandate annuitization of Superannuation Guarantee funds, which results in adverse selection that drives the pricing of private annuities for both

58

Journal of Family and Economic Issues

men and women. Women purchasers of lifetime annuities are disadvantaged further because they are priced using sex-specific actuarial factors. Inflation risk. The Swedish PAYG tier protects retirees against inflation and provides an opportunity for them to share in economic growth throughout the income range. By linking its Age Pension to average wage levels, Australia protects Age Pension beneficiaries against inflation and allows them to share in the country’s productivity growth. Inflation protection, if Superannuation accounts are annuitized, will depend on the annuity contract. Price-indexed annuities are available in the Australian market, and individuals who use their Superannuation Guarantee to purchase them receive certain tax advantages. Despite this, relatively few people have chosen this route. Widowhood. The income risk associated with widowhood arises from the loss of the husband’s income when he dies and subsequent dependence upon one’s own retirement benefits that are usually lower due to time spent out of the work force. Australia’s Age Pension shields women against work interruptions by providing benefits without regard to work history. The mandated Superannuation funds provide pension coverage to all who work but, without assured survivor benefits from their husbands’ funds, leaves them vulnerable to dependence upon their own smaller pensions. Even though Sweden’s PAYG pension does not provide mandated survivor benefits, Swedish widows benefit from mandated annuitization of their own pensions using unisex actuarial tables and most Swedish mothers will benefit from the provision of child care credits during their early years of parenting.

Conclusion Australia and Sweden are likely to continue to represent the extremes of public pension protection, even with an additional tier of protection through mandated pension coverage of Australian workers. Evidence may show that pension coverage that employers are required to pay can achieve the same goals as that offered through a universal public plan but the ultimate impact on pensionners’ incomes will be determined by the regulations governing the plans and the form of payout. Indeed, how to regulate individual accounts may

Janet Spratlin and Karen C. Holden

59

be the central argument in the debate about privatizing public insurance programs. Australia and Sweden represent the opposites in that debate. While Australia has mandated another layer of protection on its Age Pension, it continues to place full risk of investment decisions on the individual. Sweden continues its tradition of the government providing comprehensive insurance against economic risks. Reforms to its earnings-related tier and eliminating survivor benefits have shifted responsibility for retirement savings and a larger share of market risk to individual workers. At the same time, the largest component of the earnings-related tier is an account that remains managed by government and guaranteed to increase with a well-defined measure of economic growth. Mandated annuitization eliminates the risk of living longer than one’s resources can support, while the ability for individuals to adjust withdrawal rates during early retirement years provides a mechanism for couples to in effect transfer some income to the spouse (or registered partner) who is expected to live the longest. A move to more strictly earnings-based accounts, as now provided in Australia and Sweden, means that women and men will have smaller accounts when earnings are reduced due to child-rearing. However, Sweden demonstrates that even defined-contribution accounts can include redistributive features. The payment of contribution into accounts of parents targets a specific, socially valued activity and raises benefits above what they would otherwise have been. Even in earnings-based systems, adjustments can be made for activities a society wishes to subsidize. Regulating required annuitization, encouraging price-adjusted annuities, and offering various annuity arrangements also will protect women from the risks associated with longer lives.

Notes 1.

This discussion will be in terms of women as wives, mothers, and widows. However, all statements apply as well to men and husbands in comparable income positions and child-raising roles. Although most developed countries have eliminated major gender differences in benefit eligibility, some systems still retain gender differentials in benefit eligibility, usually rationalized on the basis that women remain more likely to be in need of redistribution of benefits or need to receive them earlier

60

2.

3.

4.

5. 6.

7.

8.

9.

10.

Journal of Family and Economic Issues

because earlier child-care or other home duties conflicted with a life-time full-time work career. The U.S. poverty threshold is an absolute, fixed dollar threshold that rises over time only with a change in prices. In contrast, European countries tend to use a relative standard, usually defining poverty as having an income below some specified percentage of national median disposable income. Relative poverty measures depend both upon the standard selected and the equivalence scale used. Whiteford and Kennedy (1995) show that Australia fares much better comparatively with a lower threshold but reports much higher poverty rates as that threshold rises. Shaver (1996) compares two equivalence scales, finding the relative position of Australia compared to five other countries changes with the equivalence scale used. Even in the U.S. where there isn’t support for parenting as in Sweden, unmarried women caring for children under the age of three are no more likely to work or work part-time than are their married counterparts (Bureau of Labor Statistics, 1999). For example, the U.S. Social Security System excludes the 5 lowest earning years. Sweden’s pre-reform retirement system paid benefits on the basis of the 15 highest earnings years. In the U.S., employer-provided pensions must use unisex tables in the calculation of worker and survivor benefits. Thus, an identical work record will lead to identical benefits for a male or female worker and to identical survivor benefits. This required equality is due to court interpretation of equal pay legislation in the U.S. Where the same legal interpretation is not in force, employer-provided pensions are likely to use sex-based tables. Given the option, people who for whatever reason anticipate that they will die earlier are more likely not to annuitize. Knowing this, the annuity providers price the annuities based upon the assumption that those purchasing them will live longer than average. Because women on average live longer than men, it is assumed that women will be more likely than men to annuitize their retirement accumulations. Hence, annuity prices in a voluntary market adjust to reflect this greater probability of paying annuities to those who are on average likely to live longer. In the U.S. the use of unisex tables would require like reductions for couples of the same age whether the survivor was the wife or husband. However, because women are usually younger than their husbands, at a given age the average reduction in a married man’s benefit will be larger than it would be for the average married woman of the same age. Because the old-age pension system in most countries pays a wagerelated periodic benefit, payment of these age-related survivor benefits is also the most frequently used means of insuring widows against old age poverty. In the United States, any divorced spouse married at least 10 years to a retired worker and that retiree’s current spouse are eligible for survivor

Janet Spratlin and Karen C. Holden

11. 12. 13. 14.

15.

16.

17.

18.

61

benefits based upon the retired worker’s covered earnings record. This generosity toward multiple spouses, without reduction in the retiree’s benefits, is not typical in countries that provide survivor benefits. Bateman and Piggot (1998) describe in more detail the process that led to the Superannuation Guarantee. Using a threshold of 50% of median income, 5.2% of elderly couples and 53.1% of women alone are classified as poor. But the difference between the guarantee for couples and widows implies a single person requires only 62.5% of the income of a two-person household. Indeed, Australian workers can take out their accumulated Superannuation Guarantee funds as lump sums as early as age 55, even though they are not eligible to draw the Age Pension until age 60 for women and age 65 for men. The Age of eligibility for Superannuation payments is being increased gradually to reach age 65 in year 2025. The age of eligibility for an Age Pension for women is being increased gradually to reach age 65 in year 2014. In the meantime, the eligibility age difference between the two programs is one pressure to require annuitization of Superannuation funds, since the “leakage” that takes place between age 55 and 65 will reduce the savings in the Age Pension that were assumed in the implementation of the Superannuation Guarantee. Retirement accumulations that are used to purchase CPI-indexed annuities are not counted in the assets test, one of two means tests applied to the Age Pension. A variety of inflation-adjusted annuitization schemes are discussed in some detail in Doyle and Piggott (1998). The Income-Related Old-Age Pensions Act was adopted by the Swedish Parliament June 8, 1998. The rules governing qualification for pension rights and persons who are covered by pension insurance became effective January 1, 1999, but pensions calculated under the new rules will not begin to be paid out until 2001. The reformed pension system will be phased-in gradually. People born prior to 1938 will receive pensions calculated under the old system. People in the transitional generation (born between 1938 and 1953) will receive part of their pension according to the rules of the old system and part according to the new rules through a 20-part gradual introduction of the new system. People born after 1953 will receive pensions calculated entirely according to the new rules. The “base amount” is a concept that is central to the calculations underlying several northern European pension systems. Typically the minimum annual pension is set at some multiple of the current base amount, which is adjusted regularly to reflect increases in wages or prices. In Sweden the “income base amount” is adjusted annually to reflect changes in average wages. Similarly, the “price base amount” is adjusted to reflect changes in the level of prices. In 1998, one income base amount equaled 36,400 Swedish crown, or approximately $US 4,550. Funds equal to 18.5% of the calculated child-care pensionable amount are transferred from this system’s budget to the pension system on behalf of the specified caregiver.

62

19.

Journal of Family and Economic Issues

The decision in the new system to tie the minimum pension to the “price base amount” (whereas in the earlier system it was increased with increases in average wages) represents a decision to maintain the “basket of goods” that can be purchased by the minimum pension income but not to maintain the level of minimum pension income at some proportion of the average wage income going to residents still in the labor force.

References Anderson, R. N. (1998). United States abridged life tables; 1996. 1996 National vital statistics reports, 47(13). Available at: http://www.cdc.gov/nchs/data/nvs47—13.pdf. Bateman, H., & Piggott, J. (1998). Mandating retirement provision: The Australian experience. Working Paper: University of New South Wales. Bureau of Labor Statistics. (1999). Employment characteristics of families: Summary. Labor force statistics from the current population survey. Available at: http:// stats.bls.gov/news.release/famee.nws.htm. Doyle, S., & Piggott, J. (2000). Mandatory annuity design: A preliminary study. Retirement Needs Framework. SOA Monograph M-RS00-1 (pp. 43–54). Schaumburg, IL: Society of Actuaries. Grad, S. (1996). Income of the population 55 or older, 1996. Washington, D.C.: Office of Research and Statistics, Social Security Administration. Available at: http:// www.ssa.gov/statistics/incpop55/1996/pop55toc.html. Gramlich, E. M. (1997). How does social security affect the economy? In E. R. Kingson & J. H. Schulz (Eds.), Social security in the 21st century (pp. 147–155). New York: Oxford University Press. Holden, K. C., & Nicholson, S. (1998). Selection of a joint-and-survivor pension. Discussion paper No. 1175-98, University of Wisconsin, Institute for Research on Poverty. Holden, K. C., & Smock, P. J. (1991). The economic costs of marital dissolution: Why do women bear a disproportionate cost? Annual Review of Sociology, 17, 51–78. Holden, K. C., & Zick, C. (1998). The economic impact of widowhood in the United States in the 1990s: The roles of social insurance and private pensions. In P. Flora, P. R. De Jong, J. Le Grand, & J. Kim (Eds.), The state of social welfare, 1997: International studies on social insurance and retirement, employment, family policy and health care (pp. 17–35). Brookfield, VT: Ashgate Publishing Company. Kenworthy, L. (1998). Do social-welfare policies reduce poverty? A cross-national assessment. Luxembourg income study working paper No. 188. Available at: http:// lissy.ceps.lu/wpapersd.htm. Lee, R., & Shripad, T. (1997). Death and taxes: Longer life, consumption, and social security. Demography 34(1), 67–81. Shaver, S. (1996). Universality and selectivity in income support: An assessment of the issues. Luxembourg income study working paper No. 145. Available at: http:// lissy.ceps.lu/wpapersd.htm. Siegenthaler, J. (1996). Poverty among single elderly women under different systems of old-age security: A comparative review. Luxembourg income study working paper No. 149. Available at: http://lissy.ceps.lu/wpapersc.htm. Smeeding, T. M. (1997). Reshuffling responsibilities in old age: The United States in a comparative perspective. Luxembourg income study working paper No. 153. Available at: http://lissy.ceps.lu/wpapersd.htm. Smeeding, T. M., & Dennis, S. (1998). Generations and the distribution of economic well-being: A cross-national view. Luxembourg income study working paper No. 173. Available at: http://lissy.ceps.lu/wpapersd.htm.

Janet Spratlin and Karen C. Holden

63

Social Security Administration. (1997). Annual statistical supplement to the Social Security Bulletin, 1997. Washington, D.C.: Office of Research, Evaluation and Statistics, Social Security Administration. Social Security Administration. (1999). Social security programs throughout the world. Washington, D.C.: Office of Research, Evaluation and Statistics, Social Security Administration. Available at: http://www.ssa.gov/statistics/ssptw99.html. Thompson, L. H., & Upp, M. M. (1997). The social insurance approach and social security. In E. R. Kingson & J. H. Schulz (Eds.), Social security in the 21st century (pp. 3–21). New York: Oxford University Press. Weaver, R. K. (1998). The politics of pensions: Lessons from abroad. In R. D. Arnold, M. J. Graetz, & A. H. Munnell (Eds.), Framing the social security debate: Values, politics, and economics (pp. 183–229). Washington, D.C.: National Academy of Social Insurance. Wise, D. A. (1997). Retirement against the demographic trend: More older people living longer, working less, and saving less. Demography 34(1), 83–95.