The Contributions of the Center for Retirement Research at Boston College: 2008–2017

Social Security Bulletin, Vol. 80 No. 1, 2020

Steven A. Sass is a research fellow at the Center for Retirement Research at Boston College.

The findings and conclusions presented in the Bulletin are those of the author and do not necessarily represent the views of the Social Security Administration.


Selected Abbreviations
CRR Center for Retirement Research
DB defined benefit
DC defined contribution
DYNASIM Dynamic Simulation of Income Model
FRA full retirement age
HRS Health and Retirement Study
IRA individual retirement account
MINT Modeling Income in the Near Term
NRRI National Retirement Risk Index
SSA Social Security Administration

The Social Security Administration (SSA) created the Retirement Research Consortium in 1998 “to expand the knowledge base upon which Social Security and retirement policy decisions are made.” The need to know more was critical, as the nation's retirement income system was rapidly transforming. The demographic transition to an older population was eroding the finances of Social Security and other government programs for the aged. Private-sector employers were shifting from providing defined benefit (DB) pensions to offering defined contribution (DC) plans, such as 401(k)s, with workers assuming primary responsibility for accumulating enough savings for retirement over their working years. In planning their retirement income, workers also had to account for steeply rising medical costs and the potential need for expensive long-term care. Major socioeconomic changes, such as rising wealth inequality, the increased labor force activity of women, and the declining prevalence of married-couple households were also affecting household financial preparations.

The Center for Retirement Research (CRR) at Boston College, in affiliation with the Brookings Institution, Syracuse University, and the Urban Institute, has expanded the knowledge base by producing roughly 200 research studies on key policy issues. The CRR also manages SSA-sponsored dissertation fellowship and junior-faculty grant programs to further enlarge the knowledge base and expand the pool of qualified researchers in the field. CRR studies have been broadly disseminated though working papers, issue briefs, blog posts, journal articles, and public-education booklets such as The Social Security Claiming Guide (Sass, Munnell, and Eschtruth 2016).

A previous Bulletin article reviewed the contributions of the CRR from 1998 to 2007, its first 10 years of existence (Sass 2009). This article covers the contributions of the CRR over its second 10 years, from 2008 to 2017. The first section reviews studies that address the Social Security program and its role in retirement security going forward. The second section does the same for private retirement saving. The third section reviews studies that address the medical risks and health expenditures facing an aging retiree population. The fourth section concludes by summarizing the CRR's key research contributions.

The Role of Social Security Going Forward

Although Social Security is the foundation of the nation's retirement income system, it is not designed to be the sole source of income for most retirees. Instead, it is intended to provide a base of support that replaces a portion of a household's preretirement earnings, with low earners receiving the highest replacement rates. CRR studies have assessed the effect of demographic transition and ongoing socioeconomic changes on the replacement rates that Social Security will provide in the future.

The Challenge of the Demographic Transition

Social Security is funded largely by a payroll tax on workers and their employers, and it pays retirement benefits for as long as recipients live. Following long-term increases in both the retirement-age population and average lifespans, only 2.3 workers will be paying payroll taxes for every Old-Age, Survivors, and Disability Insurance beneficiary by 2035, down from 3.4 at the end of the 20th century (Board of Trustees 2019, Table IV.B3). Congress, anticipating the demographic transition, enacted reforms in 1983 that increased Social Security revenues; for example, by accelerating scheduled payroll-tax increases and by taxing some benefits and returning the proceeds to the Social Security trust fund.1 The 1983 reforms also lowered benefits by incrementally raising the full retirement age (FRA), at which workers can claim unreduced benefits, for individuals born after 1937.2 Despite these changes, the program's trustees project that the trust fund reserves will be depleted in 2035, after which payroll taxes will be sufficient to fund only 75–80 percent of scheduled benefits.

To the extent that workers with older FRAs claim benefits commensurately later, one policy option to reduce Social Security's financing shortfall while maintaining the same replacement rate (albeit over a shorter period) is to raise the FRA again. The average retirement age of men—defined as the age at which half of men do not participate in the labor force—has increased by about 2 years from its level in the mid-1990s, based on Census Bureau data. The average retirement age for women has increased even more, as women's labor force participation patterns have been converging with those of men (Munnell 2015).

CRR studies find that the 1983 changes to FRAs were not the main factor behind the rise in the average retirement age, suggesting that unless other factors lead to later retirements going forward, any further increase in FRAs might not be accompanied by a commensurate rise in retirement ages. Hou and others (2017), using a structural model of retirement timing based on Gustman and Steinmeier (2009) and data from the University of Michigan's Health and Retirement Study (HRS), find that the shift from employer provision of DB pensions to DC plans and populationwide improvements in health condition were the most important factors driving later retirement ages. The authors find that the FRA and other Social Security program changes had lesser effects.3 Rutledge, Gillis, and Webb (2015) use HRS data to examine the factors behind the increase in average retirement ages between the mid-1990s and 2010. They find that the declining prevalence of DB pensions and retiree health insurance benefits each accounts for about 1 additional year in average retirement age, and that the change in the FRA from 66 for workers born 1943–1954 to 67 for those born in 1960 or later further raised the average retirement age, but by only 0.3 years. Burtless (2013) attributes more than half of the 9 percentage-point rise from 1985 to 2010 in the labor force participation rate of men aged 60–74 to an increase in educational attainment.

The factors these studies identify as primarily responsible for the rise in the average retirement age may have largely played out. DB pensions and retiree health insurance benefits have been relatively rare in the private sector for more than a decade. Significant further improvement in the educational attainment of workers approaching retirement is also unlikely (Burtless 2013). However, the relative health of workers approaching retirement could further improve while the physical demands of work, a factor not explicitly addressed in these studies, could continue to ease.4 Rutledge, Gillis, and Webb (2015) estimate that the factors identified in their study will raise the average retirement age by an additional year over the next three decades.

Early Claiming Patterns

Although workers on average are claiming Social Security benefits later, a substantial percentage of people still claim at 62, the earliest age of eligibility. Because benefits claimed before the FRA are actuarially reduced, the maximum possible reduction is greater for those with older FRAs. For example, monthly benefits claimed at age 62 are 80 percent of the “full” benefit for a worker whose FRA is 65, 75 percent for a worker whose FRA is 66, and 70 percent for a worker whose FRA is 67.5 The varying reductions mean that Social Security will provide less income relative to preretirement earnings to the earliest claimers with older FRAs.

To be clear, the share of workers claiming benefits at 62 has sharply declined. As recently as 1996, 50 percent of men and 57 percent of women claimed at age 62. By 2017, the shares had dropped to 35 percent of men and 40 percent of women (SSA 2018). Munnell and others (2016b), however, suggest that roughly two-thirds of workers who claim at age 62 could lack the resources to maintain their standard of living in retirement even if they annuitize their financial assets.6 Notably, however, the debate on whether a substantial share of workers will actually fall short in their retirement preparation is ongoing.7

Although workers claim retirement benefits at age 62 less frequently today than they did in the mid-1990s, the share of insured workers aged 55–59 receiving Social Security disability benefits in 2015 was about 3 percentage points higher than it was in the mid-1990s (SSA 2016). Johnson, Favreault, and Mommaerts (2010) find that disability awards rise sharply among less-educated workers as they age into their 50s and early 60s, with the households of recipients of Social Security Disability Insurance, Supplemental Security Income, workers' compensation, or veterans' benefits generally experiencing a sharp decline in income and an increased incidence of poverty. The study also finds that many less-educated workers experience work-limiting health impairments as they age but fail to qualify for disability benefits. Others have difficulty finding employment. These factors result in material hardship among less-educated workers that is reduced when they reach age 62 and can claim Social Security retirement benefits (Johnson, Mermin, and Murphy 2007; Johnson and Mermin 2009).

Munnell, Sanzenbacher, and Rutledge (2015) examine the kinds of shocks that push workers out of the labor force before their planned retirement age. Interestingly, they find that financial shocks have little effect on retirement timing. Instead, health shocks are most important, followed by layoffs, then family issues and, finally, financial shocks.8

Women's Labor Market Trends and Benefits Received by Couples and Mothers

Most workers enter retirement as part of a married-couple household. Historically, Social Security spousal and survivor benefits significantly increased the retirement incomes that married couples received. Spousal and survivor benefits were created in 1939, soon after the program's inception, when most households entering retirement were one-earner couples. The spousal benefit guarantees the lower earner (traditionally the wife) a monthly payment equal to one-half of the amount to which the higher earner would be entitled if the claim were filed at FRA (regardless of when the higher earner actually claims); the survivor benefit guarantees the widow(er) a monthly payment equal to the higher earner's actual benefit.

If the lower earner qualifies for a Social Security retired-worker benefit but the benefit amount is lower than the spousal benefit, he or she receives a “top-up” to bring the payment to the spousal benefit amount. Such a beneficiary is considered “dually entitled” because he or she receives both the worker's benefit on his or her own earnings record and the spousal top-up.

SSA actuaries estimate that retired-worker benefits claimed at FRA replace about 40 percent of a typical worker's preretirement earnings.9 Thus, for a one-earner couple, if the earner claimed at FRA, the household's benefits replace about 60 percent of the worker's preretirement earnings (because the spousal benefit is equal to one-half the earner's benefit). If the sole earner dies, the survivor benefit equals two-thirds of the combined amount the couple had received, which approximates an “equivalent” benefit for a single individual under standard equivalence scales (Forster and Levy 2013).10

Most married women today work outside the home and qualify for Social Security retirement benefits based on their own earnings record. It is also increasingly common for workers to enter retirement as part of an unmarried-couple household or as a divorced or unmarried parent who is not eligible for spousal or survivor benefits.11 CRR studies have assessed the effect of these changes on retirement income from Social Security.

The employment of married women increased dramatically in the 1960s and 1970s. This trend increased household preretirement incomes but the increase in household Social Security benefits was not commensurate. Wu and others (2013), using SSA's Modeling Income in the Near Term (MINT) microsimulation model, find that declining spousal benefits were the main contributor to a decline in average household Social Security replacement rates. Those rates dipped from 50 percent for the birth cohorts whose members entered the labor force in the 1950s to 45 percent for those who entered in the 1960s and 1970s.

Although spousal benefits are declining in importance (both as a share of total recipients and as a share of the total benefits received by those who are dually entitled), survivor benefits remain important, as married women generally continue to earn less than their husbands. However, Butrica and Smith (2012a), using MINT, estimate that one-third of married women who entered the labor force in the 1990s will earn as much as or more than their husbands, compared with 18 percent of married women who entered the labor force in the 1960s. As a result, these women will not receive survivor benefits and their household Social Security income will decline more after their husbands' death than it did for previous generations. For example, assuming that both spouses claim at their FRA, the average widow in the 1966–1975 birth cohort will get only 54 percent of what the couple received when the husband was still alive, down from 59 percent for widows born during 1936–1945.

Another study by Butrica and Smith (2012b) using MINT projects an increase in the share of mothers entering retirement who never married or are divorced from a marriage that lasted fewer than 10 years—from 7 percent of war baby mothers (born 1936–1945) to 15 percent of generation X mothers (born 1966–1975).12 Although these women are not entitled to spousal or survivor benefits, child-rearing responsibilities are likely to have limited the time they spent in the labor market, impeded their advancement to better-paying jobs, and limited their potential Social Security retired-worker benefits. Rutledge, Zulkarnain, and King (2017), using data from the HRS and linked Social Security administrative data, find that the benefits of mothers are 16 percent less than the benefits of women with no children, and each additional child reduces benefits by about another 3 percent. Although widows currently constitute the largest poverty population among the aged, divorced and never-married mothers have much higher poverty rates and are projected to account for increasing shares of the aged population (Johnson, Favreault, and Goldwyn 2003).

Potential Changes to Social Security

The CRR studies reviewed above address how current programmatic, claiming, and labor force patterns might affect household Social Security income levels relative to preretirement earnings. Going forward, Social Security faces a long-term financing shortfall. To address it, experts have presented many policy proposals. One example, the bipartisan National Commission on Fiscal Responsibility and Reform's 2010 Bowles-Simpson proposal, would reduce the FRA benefit for higher earners; raise the FRA and earliest age of eligibility in line with rising longevity, while exempting low earners; and provide an enhanced minimum benefit for long-career low-wage workers.13

Using the Urban Institute's Dynamic Simulation of Income Model (DYNASIM), Favreault and Steuerle (2012) assess how Bowles-Simpson could affect poverty and other key outcomes for the aged. Despite the proposal's protections for low earners, the authors find that poverty would increase relative to levels projected based on currently scheduled benefits.14 The study also points out that Bowles-Simpson does not address issues raised by changing marital patterns beyond the proposal's protections for all low-income beneficiaries.15

The CRR study highlights the tradeoffs that policymakers face in addressing Social Security's long-term financing shortfall while continuing to provide retirement income security. Bowles-Simpson and similar proposals highlight the importance of minimizing old-age poverty and preserving benefits for low earners. To date, program responses to demographic transitions, changing marital patterns, and increased income inequality have suggested that Social Security is likely to play a reduced role for workers who retire early and for married couples and widow(er)s who will increasingly enter retirement without spousal or survivor benefits. If Social Security adopted reforms similar to those proposed by Bowles-Simpson, benefits would be lower—relative to preretirement earnings—than in the past, especially for middle and higher earners. If the role of Social Security changes, retirees may increasingly depend on income provided by their own retirement savings.

The Role of Retirement Savings Going Forward

Besides Social Security, the major source of retirement income is savings accumulated over the course of a working career. Such savings include financial assets (which for most retirees consist primarily of employer-sponsored DC plans) and home equity (which can be tapped to cover living expenses).

In recent decades, most private-sector employers have switched from providing DB pensions to offering DC plans to their employees.16 The transition, which began in the 1980s, is now largely complete. Although many workers entering retirement today still have DB pension benefits that accrued earlier in their careers, DB pensions are declining as a source of retirement income.

Munnell and others (2016a) use data from the HRS to assess the effect of the DB-to-DC transition on the household pension wealth of workers approaching retirement. Comparing the combined amount of DC plan and individual retirement account (IRA) balances with DB pension accruals, the authors find that DC/IRA balances accounted for 35 percent of the pension wealth of households headed by individuals aged 51–56 in 1992, with DB pension accruals accounting for the other 65 percent. In 2010, the respective shares were 62 percent and 38 percent. Although average and median pension wealth remained much the same across the period, the share of households with no pension wealth had increased, and high-income households held a greater share of pension wealth in 2010. Assuming households annuitized their savings at retirement, the share of preretirement earnings that was replaced by the resulting income stream was smaller in 2010 than in 1992, as households in 2010 had higher preretirement earnings, would have to use commercially available annuities (which provide less income from a given amount of savings than would a DB plan), and were adversely affected by declining interest rates.17

The Pension Protection Act of 2006 facilitated the widespread adoption of an important 401(k) plan innovation: automatic enrollment for new employees, increasingly accompanied by automatic annual contribution escalation, with the employee able to opt out of either feature.18 Automatic enrollment with automatic contribution escalation tends to significantly increase the participation and contribution rates of young and low-income workers (Madrian and Shea 2001; Madrian 2012).19 Chetty and others (2013) use detailed Danish data on household finances and confirm that workers who were “nudged” into contributing more to their retirement plans did not make offsetting reductions in other types of saving. The authors also find that workers generally do not respond to changes in retirement-saving tax incentives, which require an active decision to change how much they save. The study concludes that automatic enrollment can increase a worker's retirement saving much more than government tax subsidies can, and do so at much lower cost.20

Butrica and Karamcheva (2012) and Soto and Butrica (2009) report that employers too have found automatic enrollment to be more effective than financial incentives at influencing worker retirement saving.21 Butrica and Karamcheva, using data from the National Compensation Survey, find that plans with automatic enrollment have higher participation rates with relatively low default contributions and employer-match rates. These results suggest that behavioral nudges such as automatic enrollment can increase plan participation while potentially allowing firms to keep their employee benefit costs roughly constant.22 In this situation, employees who previously did not participate could end up saving more while others might save less because of the reduced default rate and match.

Another major component of wealth for many retirees is home equity. More than 80 percent of retirees own their own homes and home equity is a larger store of savings than financial assets for most low- and moderate-income households. The Federal Housing Administration's Home Equity Conversion Mortgage (HECM) program provides reverse mortgage loans for homeowners aged 62 or older.23 HECM allows households to tap their home equity, without requiring any repayments, for as long as they live in the house. Mudrazija and Butrica (2017) estimate that the HECM program could raise the income of an aged homeowner at the median by more than one-third, and reduce the share of homeowners with incomes below 50 percent of median household income by 6–7 percentage points, leaving just 13–14 percent under that low-income benchmark.24

Projecting Retirement Saving and Preparedness

Butrica, Smith, and Iams (2012) and Johnson and others (2017) use MINT to project retirement incomes assuming households annuitize 80 percent of their financial savings. Both studies find that retirement incomes will replace a declining share of preretirement incomes of all households, particularly among higher- and middle-income households, as retirement savings are not projected to offset reduced replacement rates from Social Security.25

Munnell, Rutledge, and Webb (2014) seek to reconcile differences between conflicting estimates of retirement preparedness. The authors compare estimates from the CRR's National Retirement Risk Index (NRRI) with others from a Michigan Retirement Research Center study (Scholz and Seshadri 2008) based on HRS data and a model of optimal lifetime consumption. The NRRI is constructed using data from the Survey of Consumer Finances. It projects replacement rates for current working-age households at retirement, assuming they will retire at age 65 and annuitize all their DC/IRA savings and the proceeds of a reverse mortgage. The authors compare the projected replacement rates with target replacement rates to calculate the share of households that are at risk of inadequate retirement income. The authors estimate that in 2004, 35 percent of households aged 50–58 were at risk.

Scholz and Seshadri (2008) find more optimistic results. They estimate that only 8 percent of households in their 50s in 2004 had not saved adequately for retirement. Munnell, Rutledge, and Webb (2014) find that the difference between the Scholz and Seshadri and the NRRI assessments is largely due to two assumptions. First, unlike the NRRI, Scholz and Seshadri assume that households choose a declining consumption path in retirement—that those households devote a smaller share of their resources to support consumption as they age and the likelihood of survival declines. Second, Scholz and Seshadri assume that when children leave home, the household saves the income it formerly spent on the children. The NRRI assumes that household saving remains the same, which increases the income spent on the parents' consumption and raises their standard of living. The Scholz and Seshadri assumption results in a lower target retirement income and more savings for meeting that target. However, Dushi and others (2015), using data from the HRS and the Survey of Income and Program Participation, show that household saving does not change much when children leave home.26

Changes That Could Facilitate More Retirement Saving

The studies reviewed above indicate that future retirees cannot expect income derived from savings to offset the projected declines in household Social Security benefit replacement rates and the diminution of DB pension income. To assure future retirees of income levels similar to those of current retirees, workers will need to save more, take on more risk with their assets, work longer, or draw more income out of their savings than current retirees have had to do.

Further behavioral or informational innovations could increase household retirement saving. For example, some states have introduced automatic IRA-enrollment programs, which require all employers above specified sizes who do not offer a DB or DC plan to enroll their workers in a payroll-deduction IRA, from which workers can then opt out. Butrica and Smith (2016), using DYNASIM to generate preimplementation estimates of the program's potential effect, find that such initiatives might only modestly increase retirement saving. They find that program design factors, including the number of firms covered, contribution limits, default contribution rates, and investment options will determine the extent of its effect. To date, the early-implementation states have adopted some design features associated with the higher participation and saving estimates in the Butrica and Smith model. For example, Oregon covers all employers, has a 5-percent default contribution rate that automatically escalates in subsequent years to 10 percent, and designates a target date fund as the default investment choice. In initial results from Oregon's program, the majority of eligible workers participate and generally stick with the default contribution rate. One challenge facing Oregon and other auto-IRA states appears to be helping employers who are unfamiliar with the program to provide timely and accurate data and to process payroll deductions (Belbase and Sanzenbacher 2018).27

By itself, accumulating savings does not solve the problem of preparing for retirement. As employer pension offerings have switched from DB plans to DC plans, the need for retirees to draw an adequate income out of their savings has become more acute. Projections of retirement income adequacy assume that retirees either annuitize their savings or put them to use under an optimal drawdown strategy; but retirees generally do neither (Poterba, Venti, and Wise 2011). Pashchenko (2013) identifies various economic reasons retirees may tend not to follow these assumptions, such as bequest motives, medical expense uncertainty, and the illiquidity of housing wealth. Reviews of the Retirement Research Consortium literature (Sass 2016b; 2017) identify powerful behavioral and informational factors that impede retirees from drawing down their financial assets and home equity, which would allow them to increase current consumption. However, pressures to tap their savings will be greater going forward than they have been in the past.

Medical Risks Going Forward

The aged need more medical care than younger people do, and for many years, the cost of care has risen faster than wages, gross domestic product, and retirement incomes. That trend is expected to continue. The need for long-term services and supports to compensate for age-related declines in physical and mental abilities will also spike when the large baby boomer cohorts reach advanced old age. Studies conducted by the CRR and its affiliates assess the magnitude of these challenges and how they might be addressed.

Rising Costs

Because of rising medical costs, the federal government created Medicare and Medicaid in 1965 to help the aged and other vulnerable populations. As the cost of care continued to rise, the programs expanded. Medicare covers about 80 percent of the hospital and physician costs of retirees and, since 2006, it has offered prescription drug coverage. Medicaid insures the very-low-income aged, covers the cost of long-term care for retirees who cannot afford it, and since 1990, covers the Medicare copayments and premiums for low-income retirees with incomes above the Medicaid eligibility limits.28

Even though Medicare and Medicaid bear most of the cost, retirees spend a large and growing share of their incomes on medical care. For example, McInerney, Rutledge, and King (2017), using HRS data, find that median-income aged households spend nearly one-quarter of their Social Security benefits and more than 10 percent of their incomes on medical care, and project that those shares will increase over the next decade. Favreault (2015), using DYNASIM, projects that medical expenditures will require more than one-half of the Social Security benefits and nearly 20 percent of the total incomes of median-income aged households by 2055.29 For lower-income households not eligible for Medicaid and households with above-average medical needs, the projected burden is significantly greater.

The Burdens of Advanced Old Age

The incidence of physical and mental impairments rises sharply after age 85. As a result, individuals in this age group often cannot perform activities of daily living (such as eating, dressing, and bathing) and instrumental activities of daily living (such as shopping, cooking, and managing money) without assistance. The oldest baby boomers will cross this age threshold in 2031, and the share of the population that is aged 85 or older will rise rapidly until it stabilizes—at a much higher level—by the middle of the century.

CRR studies address key challenges in negotiating advanced aging. For example, Belbase and Sanzenbacher (2016; 2017) assess the effect of the high incidence of dementia in advanced old age on the management of household finances. Using data from the Johns Hopkins University's National Health and Aging Trends Study, the authors find that 85 percent of respondents with dementia have family members—typically, adult children and nonimpaired spouses—to help them manage their finances, primarily in paying bills. Representative payees, guardians, or other formally appointed caretakers assist another 10 percent, many of whom have no family members living nearby. The study finds that such assistance is effective. Among those who receive help, the incidence of financial hardship is much the same as that for otherwise similar individuals without dementia—and is much lower than that for the 5 percent of respondents with dementia who do not get any assistance.

The Effect on Government Budgets

Rising medical costs and declining family supports will stress government budgets as well as the budgets of aged households. The federal government covers 80 percent of an aged household's Medicare costs and federal and state governments cover much of the cost of Medicaid benefits. Because the median age of the population will continue to rise, governments will need to support many more aged households with medical and long-term care needs.

In the past, Congress expanded the Medicare and Medicaid programs in response to the rising cost of care. It added optional prescription drug coverage to Medicare and had Medicaid cover the Medicare premiums and copayments of low-income aged households that were not eligible for means-tested Supplemental Security Income payments. If medical costs rise as projected, policymakers might take additional action to address the cost burden to households.30

Favreault (2015) estimates future retiree incomes and medical expenditures under different assumptions, including an across-the-board cut in Social Security benefits in response to the program's financing shortfall and a more rapid increase in medical costs. If Social Security benefits are cut, more retirees will qualify for Medicaid. A more rapid increase in medical costs will further stress government budgets. The study's projections indicate that the ability of future retirees to manage out-of-pocket health costs will be substantially affected by cost growth, Medicare and Medicaid spending, and the income levels provided by Social Security and private retirement-income plans.


Twenty years have passed since the SSA created the Retirement Research Consortium, and the demographic, economic, and programmatic forces transforming the nation's retirement income system are now largely in place. We are entering the final phase of the long upward trend in the nation's median age. The shift from employer-provided DB pensions to DC plans in the private sector is largely complete. Marital-status trends have changed and employment among women has increased. Changes in the distribution of income and wealth are hard to predict, and inequality might stabilize, decline, or further expand. The one trend expected to continue is the cost of medical care rising faster than household incomes.

The CRR studies reviewed here suggest that retirees and governments will likely face challenges in maintaining the level of income security that retirees experienced in the past. Now, the nation faces a critical and action-forcing event: The Social Security trust fund reserves currently are projected to be depleted in 2035, only 15 years away. At that point, Social Security will be able to pay only 75–80 percent of scheduled benefits—to existing and new beneficiaries. In that scenario, households entering retirement will have lower benefits, little or no DB pension income, and higher out-of-pocket medical costs.

No consensus yet exists on how to respond to the key challenge facing the nation's retirement income system: the finances of the government's Social Security, Medicare, and Medicaid programs. The CRR has helped “expand the knowledge base upon which Social Security and retirement policy decisions are made” by contributing studies on longevity, interactions between the government's medical insurance and cash benefit programs for the aged, and the likely effect of changing family structures on the demand for government-funded services when the baby boomers enter advanced old age. Changes in fertility rates, given women's changing economic roles, could also significantly affect the long-term solvency of government programs for the aged. As the depletion of the Social Security trust fund reserves draws nearer, these issues will be increasingly important research priorities.


1 Taxation of benefits applies to beneficiaries with incomes above specified nominal dollar amounts.

2 The FRA remained 65 for workers born in 1937 or earlier. The new FRAs rose in 2-month increments for members of each successive birth cohort from 1938 through 1942, became 66 for members of the 1943–1954 birth cohorts, again rose in 2-month increments for members of each successive birth cohort from 1955 through 1959, and became 67 for workers born in 1960 or later.

3 Other Social Security changes that encouraged later retirement included raising the existing delayed retirement credits, which provide higher monthly benefits if claimed after FRA; and eliminating the retirement earnings test for beneficiaries who work after attaining their FRA.

4 Physical job demands and declining physical abilities are commonly cited as resulting in the “superannuation” of some older blue-collar workers. In addition, Belbase and others (2015) and Belbase, Sanzenbacher, and Gillis (2015) find that cognitive job demands and declining cognitive abilities result in the superannuation of some older white-collar workers. As blue-collar employment declines and white-collar employment expands, cognitive job demands and declining cognitive abilities will become more important. Belbase, Sanzenbacher, and Gillis nevertheless find that older blue-collar workers are especially susceptible to superannuation.

5 Because the age-67 FRA applies to workers born in 1960 or later, the first year such workers will be eligible to claim is 2022.

6 The authors estimate that the share of age-62 claimers who were unable to maintain their preretirement standard of living rose from 60 percent of those who reached age 62 during 1993–1998 to 66 percent of those who reached age 62 during 2004–2009. Age-62 claimers who had the resources to maintain their standard of living were better educated, had higher incomes and wealth, and were much more likely to have employer-provided DB pensions and retiree health insurance than those who did not.

7 See, for example, the discussion in Munnell, Rutledge, and Webb (2014).

8 Financial shocks are highly correlated with labor market shocks, and have offsetting effects on retirement behavior: Downturns increase a worker's desire to remain employed and delay retirement, but also reduce employer demand for workers.

9 Specifically, Clingman, Burkhalter, and Chaplain (2019) estimate that the benefit of a worker with scaled medium earnings who retires at FRA in 2019 replaces 41.0 percent of preretirement earnings.

10 The survivor benefit initially guaranteed the widow(er) 75 percent of the higher earner's benefit, and was increased to 100 percent in 1972. Survivors are also guaranteed 82.5 percent of the higher earner's FRA benefit, if greater than the higher earner's actual benefit. Both spousal and survivor benefits are reduced if claimed prior to the recipient's FRA.

11 Divorced individuals are entitled to spousal and survivor benefits based on their ex-spouse's earnings record if the marriage lasted 10 years or more.

12 Specifically, the study finds that 10 percent of generation X mothers will never have been married, versus 3 percent of war baby mothers; and that 5 percent of generation X mothers will have been divorced from a marriage of fewer than 10 years, versus 4 percent of war baby mothers.

13 Bowles-Simpson would also adjust the maximum taxable-income level so that 90 percent of U.S. earnings would be subject to payroll taxes; adopt an inflation measure that would reduce Social Security cost-of-living increases, which would be offset by an increase in benefits for long-term beneficiaries; and require all new state and local government workers to participate in the program. Exempting low-wage workers from the FRA and early-eligibility age increases was justified by the fact that longevity has risen primarily for high-wage workers, and less so for low-wage workers, as documented in CRR studies (Bosworth and Burke 2014; Bosworth, Burtless, and Zhang 2015; Bosworth and Zhang 2015; Sanzenbacher and others 2017; Sanzenbacher and Ramos-Mercado 2016).

14 The authors suggest that one reason for the projected increase in aged poverty is that the proposal bases its minimum benefit and hardship exemptions on career lengths and thus offers less protection for the chronically poor.

15 The study focuses on the proposal's failure to address issues of horizontal equity, or differences in the ratio of household benefits to household taxes, in the current spousal and survivor benefit design. The study also assesses Bowles-Simpson incentives for workers to delay retirement relative to the current incentives and “reasonably feasible” alternatives.

16 Most public-sector workers, who comprise about 15 percent of the total nonfarm workforce (Bureau of Labor Statistics 2019), still have DB plans.

17 The study combines IRA balances and DC assets because most IRA assets are 401(k) plan rollovers. It uses standard methods (developed in Gustman, Steinmeier, and Tabatabai 2010) for valuing DB accruals and the authors estimate the income from DC-plan savings by annuitizing projected account balances while accounting for the higher retirement ages of more recent cohorts and declining interest rates. Sorokina, Webb, and Muldoon (2008); Munnell, Aubry, and Crawford (2015); and Karamcheva and Sanzenbacher (2010) also study the effects of the DB-to-DC shift on trends in pension holdings and retirement savings.

18 The act provided various safe harbors for employers implementing automatic enrollment, such as default investment allocations, and affirmed that state payroll withholding laws applied to 401(k) defaults (Beshears and others 2010).

19 Automatic-enrollment plans have a default investment vehicle, typically a “target date fund,” which usually performs better than what workers would otherwise elect. Munnell, Orlova, and Webb (2012) discuss the limited importance of asset allocation and Kopcke and others (2013) discuss uncertainty in determining optimal asset allocations.

20 Note that the Danish and U.S. tax systems differ and, because of declining marginal tax rates and changes in the tax treatment of investment income in the United States, the value of government tax preferences for retirement saving has significantly declined (Burtless and Toder 2010). For this and other reasons, the findings for Denmark may not necessarily translate to the United States.

21 One possible explanation is that behavioral defaults enable employers to meet antidiscrimination tests less expensively than matching contributions do (Sass 2016a).

22 Default contributions are typically much lower than contributions to traditional plans, which tend to gravitate toward the maximum level that the employer matches. However, employers are increasingly adopting automatic escalation of the default contribution amount (Vanguard 2017).

23 Munnell and Sass (2014) discuss the HECM program as revised in 2013. The program was further revised in 2017.

24 The results are based on the HECM program prior to the 2017 revision. Wimer and Manfield (2015) also consider the role of reverse mortgages in estimating potential income available to aged households.

25 Sass (2018b) also assesses the resource adequacy of future retirees.

26 Coe and Webb (2010) study household consumption (rather than saving) among empty nesters.

27 Sass (2014) and Cribb and Emmerson (2019) describe a universal automatic-enrollment program with both employer and government matching contributions, recently rolled out in the United Kingdom. That program should be more effective in raising retirement saving, but at a higher cost to the government.

28 Specifically, Medicaid covers the Medicare premiums and copayments of aged individuals with income below the federal poverty level and the Medicare premiums of many retirees with incomes up to 135 percent of the federal poverty level (Sass 2018a).

29 The model uses data from the HRS and the Centers for Medicare and Medicaid Services' Medicare Current Beneficiary Survey. The baseline model assumes that Social Security benefits are not reduced, retirees annuitize their financial assets, medical costs rise as projected in the Medicare Trustees' baseline estimate, and medical expenditures and supplementary insurance purchases respond to changes in medical costs and household incomes.

30 Sass (2018a) summarizes the trends that point toward the expansion of Medicaid; Butrica, Murphy, and Zedlewski (2010) examine the effect of medical expenditures on the projected incidence of poverty among aged households.


Belbase, Anek, Mashfiqur R. Khan, Alicia H. Munnell, and Anthony Webb. 2015. “Slowed or Sidelined? The Effect of 'Normal' Cognitive Decline on Job Performance Among the Elderly.” Working Paper No. 2015-12. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Belbase, Anek, and Geoffrey T. Sanzenbacher. 2016. “Cognitive Impairment and Social Security's Representative Payee Program.” Working Paper No. 2016-12. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2017. “Dementia, Help with Financial Management, and Well-Being.” Working Paper No. 2017-11. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2018. “How Have Workers Responded to Oregon's Auto-IRA?” Issue in Brief No. 18-22. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Belbase, Anek, Geoffrey T. Sanzenbacher, and Christopher M. Gillis. 2015. “Does Age-Related Decline in Ability Correspond with Retirement Age?” Working Paper No. 2015-24. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Beshears, John, James J. Choi, David Laibson, Brigitte C. Madrian, and Brian Weller. 2010. “Public Policy and Saving for Retirement: The 'Autosave' Features of the Pension Protection Act of 2006.” In Better Living Through Economics: How Economic Research Improves Our Lives, edited by John J. Siegfried (274–290). Cambridge, MA: Harvard University Press.

[Board of Trustees] Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds. 2019. The 2019 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and the Federal Disability Insurance Trust Funds. Washington, DC: Government Publishing Office.

Bosworth, Barry P., and Kathleen Burke. 2014. “Differential Mortality and Retirement Benefits in the Health and Retirement Study.” Working Paper No. 2014-4. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Bosworth, Barry P., Gary Burtless, and Kan Zhang. 2015. “Sources of Increasing Differential Mortality Among the Aged by Socioeconomic Status.” Working Paper No. 2015-10. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Bosworth, Barry P., and Kan Zhang. 2015. “Evidence of Increasing Differential Mortality: A Comparison of the HRS and SIPP.” Working Paper No. 2015-13. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Bureau of Labor Statistics. 2019. “Employment Projections: Employment By Major Industry Sector.”

Burtless, Gary. 2013. “Can Educational Attainment Explain the Rise in Labor Force Participation at Older Ages?” Issue in Brief No. 13-13. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Burtless, Gary, and Eric Toder. 2010. “The Shrinking Tax Preference for Pension Savings: An Analysis of Income Tax Changes, 1985–2007.” Working Paper No. 2010-3. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Butrica, Barbara A., and Nadia S. Karamcheva. 2012. “Automatic Enrollment, Employee Compensation, and Retirement Security.” Working Paper No. 2012-25. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Butrica, Barbara A., Daniel P. Murphy, and Sheila R. Zedlewski. 2010. “How Many Struggle to Get by in Retirement?” The Gerontologist 50(4): 482–494.

Butrica, Barbara A., and Karen E. Smith. 2012a. “The Impact of Changes in Couples' Earnings on Married Women's Social Security Benefits.” Social Security Bulletin 72(1): 1–9.

———. 2012b. “The Retirement Prospects of Divorced Women.” Social Security Bulletin 72(1): 11–22.

———. 2016. “The Potential Impact of Mandated Employer Pension Programs on Retirement Savings.” Paper presented at the 18th Annual Meeting of the Retirement Research Consortium, Washington, DC (August 4–5).

Butrica, Barbara A., Karen E. Smith, and Howard M. Iams. 2012. “This Is Not Your Parents' Retirement: Comparing Retirement Income Across Generations.” Social Security Bulletin 72(1): 37–58.

Chetty, Raj, John N. Friedman, Soren Leth-Petersen, Torben Heien Nielsen, and Tore Olsen. 2013. “Subsidies vs. Nudges: Which Policies Increase Saving the Most?” Issue in Brief No. 13-3. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Clingman, Michael, Kyle Burkhalter, and Chris Chaplain. 2019. “Replacement Rates for Hypothetical Retired Workers.” Actuarial Note No. 2019.9. Baltimore, MD: SSA, Office of the Chief Actuary.

Coe, Norma B., and Anthony Webb. 2010. “Children and Household Utility: Evidence from Kids Flying the Coop.” Working Paper No. 2010-16. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Cribb, Jonathan, and Carl Emmerson. 2019. “Requiring Auto-Enrollment: Lessons from UK Retirement Plans.” Issue in Brief No. 19-6. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Dushi, Irena, Alicia H. Munnell, Geoffrey T. Sanzenbacher, Anthony Webb, and Anqi Chen. 2015. “Do Households Increase Their Savings When the Kids Leave Home?” Working Paper No. 2015-26. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Favreault, Melissa M. 2015. “How Would Social Security Changes Affect Medicare Costs and Seniors' Out-of-Pocket Spending?” Paper presented at the 17th Annual Meeting of the Retirement Research Consortium, Washington, DC (August 6–7).

Favreault, Melissa M., and C. Eugene Steuerle. 2012. “Measuring Social Security Proposals by More than Solvency: Impacts on Poverty, Progressivity, Horizontal Equity, and Work Incentives.” Working Paper No. 2012-15. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Forster, Michael, and Horacio Levy. 2013. “The OECD Approach to Measure and Monitor Income Poverty Across Countries.” Working Paper No. 17. Geneva: United Nations Economic Commission for Europe, Conference of European Statisticians.

Gustman, Alan L., and Thomas Steinmeier. 2009. “How Changes in Social Security Affect Recent Retirement Trends.” Research on Aging 31(2): 261–290.

Gustman, Alan L., Thomas L. Steinmeier, and Nahid Tabatabai. 2010. Pensions in the Health and Retirement Study. Cambridge, MA: Harvard University Press.

Hou, Wenliang, Alicia H. Munnell, Geoffrey T. Sanzenbacher, and Yinji Li. 2017. “Why Are U.S. Households Claiming Social Security Later?” Working Paper No. 2017-3. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Johnson, Richard W., Melissa M. Favreault, and Joshua H. Goldwyn. 2003. “Employment, Social Security, and Future Retirement Outcomes for Single Mothers.” Working Paper No. 2003-14. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Johnson, Richard W., Melissa M. Favreault, and Corina Mommaerts. 2010. “Work Ability and the Social Insurance Safety Net in the Years Prior to Retirement.” Working Paper No. 2009-28. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Johnson, Richard W., and Gordon Mermin. 2009. “Financial Hardship Before and After Social Security's Early Eligibility Age.” Working Paper No. 2009-8. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Johnson, Richard W., Gordon Mermin, and Dan Murphy. 2007. “The Impact of Late-Career Health and Employment Shocks on Social Security and Other Wealth.” Working Paper No. 2007-26. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Johnson, Richard W., Karen E. Smith, Damir Cosic, and Claire Xiaozhi Wang. 2017. “Retirement Prospects for Millennials: What Is the Early Prognosis?” Working Paper No. 2017-17. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Karamcheva, Nadia S., and Geoffrey T. Sanzenbacher. 2010. “Is Pension Inequality Growing?” Issue in Brief No. 10-1. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Kopcke, Richard W., Anthony Webb, Joshua Hurwitz, and Zhenyu Li. 2013. “Rethinking Optimal Wealth Accumulation and Decumulation Strategies in the Wake of the Financial Crisis.” Working Paper No. 2013-1. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Madrian, Brigitte C. 2012. “Matching Contributions and Savings Outcomes: A Behavioral Economics Perspective.” In Matching Contributions for Pensions: A Review of International Experience, edited by Richard Hinz, Robert Holzmann, David Tuesta, and Noriyuki Takayama (289–310). Washington, DC: World Bank.

Madrian, Brigitte C., and Dennis F. Shea. 2001. “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior.” Quarterly Journal of Economics 116(4): 1149–1187.

McInerney, Melissa, Matthew S. Rutledge, and Sara Ellen King. 2017. “How Much Does Out-of-Pocket Medical Spending Eat Away at Retirement Income?” Working Paper No. 2017-13. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Mudrazija, Stipica, and Barbara A. Butrica. 2017. “Homeownership, Social Insurance, and Old-Age Security in the United States and Europe.” Working Paper No. 2017-15. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H. 2015. “The Average Retirement Age—An Update.” Issue in Brief No. 15-4. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Jean-Pierre Aubry, and Caroline V. Crawford. 2015. “How Has Shift to Defined Contribution Plans Affected Saving?” Issue in Brief No. 15-16. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Wenliang Hou, Anthony Webb, and Yinji Li. 2016a. “Pension Participation, Wealth, and Income: 1992–2010.” Working Paper No. 2016-3. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Natalia Sergeyevna Orlova, and Anthony Webb. 2012. “How Important Is Asset Allocation to Financial Security in Retirement?” Working Paper No. 2012-13. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Matthew S. Rutledge, and Anthony Webb. 2014. “Are Retirees Falling Short? Reconciling the Conflicting Evidence.” Working Paper No. 2014-16. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Geoffrey T. Sanzenbacher, and Matthew S. Rutledge. 2015. “What Causes Workers to Retire Before They Plan?” Working Paper No. 2015-22. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., Geoffrey T. Sanzenbacher, Anthony Webb, and Christopher M. Gillis. 2016b. “Are Early Claimers Making a Mistake?” Working Paper No. 2016-5. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Munnell, Alicia H., and Steven A. Sass. 2014. “The Government's Redesigned Reverse Mortgage Program.” Issue in Brief No. 14-1. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Pashchenko, Svetlana. 2013. “Accounting for Non-Annuitization.” Journal of Public Economics 98: 53–67.

Poterba, James M., Steven F. Venti, and David A. Wise. 2011. “The Drawdown of Personal Retirement Assets.” NBER Working Paper 16675. Cambridge, MA: National Bureau of Economic Research.

Rutledge, Matthew S., Christopher M. Gillis, and Anthony Webb. 2015. “Will the Average Retirement Age Continue to Increase?” Working Paper No. 2015-16. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Rutledge, Matthew S., Alice Zulkarnain, and Sara Ellen King. 2017. “How Much Does Motherhood Cost Women in Social Security Benefits?” Working Paper No. 2017-14. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Sanzenbacher, Geoffrey T., and Jorge D. Ramos-Mercado. 2016. “Calculating Expected Social Security Benefits by Race, Education and Claiming Age.” Working Paper No. 2016-14. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Sanzenbacher, Geoffrey T., Anthony Webb, Candace M. Cosgrove, and Natalia Orlova. 2017. “Rising Inequality in Life Expectancy by Socioeconomic Status.” Working Paper No. 2017-2. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Sass, Steven A. 2009. “The Research Contributions of the Center for Retirement Research at Boston College.” Social Security Bulletin 69(4): 35–49.

———. 2014. “The U.K.'s Ambitious New Retirement Savings Initiative.” Issue in Brief No. 14-5. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2016a. “Can We Increase Retirement Saving?” Issue in Brief No. 16-15. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2016b. “How Can We Realize the Value That Annuities Offer in a 401(k) World?” Issue in Brief No. 16-2. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2017. “Is Home Equity an Underutilized Retirement Asset?” Issue in Brief No. 17-6. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2018a. “How Medicaid Helps Older Americans.” Issue in Brief No. 18-5. Chestnut Hill, MA: Center for Retirement Research at Boston College.

———. 2018b. “Will the Financial Fragility of Retirees Increase?” Issue in Brief No. 18-4. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Sass, Steven, Alicia H. Munnell, and Andrew Eschtruth. 2016. The Social Security Claiming Guide. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Scholz, John Karl, and Ananth Seshadri. 2008. “Are All Americans Saving 'Optimally' for Retirement?” MRRC Working Paper No. 2008-189. Ann Arbor, MI: University of Michigan Retirement Research Center.

Social Security Administration. 2016. Annual Statistical Supplement to the Social Security Bulletin, 2015. Publication No. 13-11700. Washington, DC: SSA.

———. 2018. Annual Statistical Supplement to the Social Security Bulletin, 2017. Publication No. 13-11700. Washington, DC: SSA.

Sorokina, Olga, Anthony Webb, and Dan Muldoon. 2008. “Pension Wealth and Income: 1992, 1998, and 2004.” Issue in Brief No. 8-1. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Soto, Mauricio, and Barbara A. Butrica. 2009. “Will Automatic Enrollment Reduce Employer Contributions to 401(k) Plans?” Working Paper No. 2009-33. Chestnut Hill, MA: Center for Retirement Research at Boston College.

SSA. See Social Security Administration.

Vanguard. 2017. “How America Saves 2017: Vanguard 2016 Defined Contribution Plan Data.” Valley Forge, PA: The Vanguard Group, Inc.

Wimer, Christopher, and Lucas Manfield. 2015. “Elderly Poverty in the United States in the 21st Century: Exploring the Role of Assets in the Supplemental Poverty Measure.” Working Paper No. 2015-29. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Wu, April Yanyuan, Nadia S. Karamcheva, Alicia H. Munnell, and Patrick J. Purcell. 2013. “How Do Trends in Women's Labor Force Activity and Marriage Patterns Affect Social Security Replacement Rates?Social Security Bulletin 73(4): 1–24.