The Brookings Institution
By: David C. John
September 1, 2010
Good morning. I am David C. John, the senior research dellow for Retirement Security and Financial Institutions at the Heritage Foundation. I am also the deputy director of the Retirement Security Project (RSP), a joint effort of the Brookings Institution and Georgetown University supported by the Rockefeller Foundation and the Pew Charitable Trusts.
This morning, I am pleased to present parts of three papers written by former RSP colleagues and outside experts on specific retirement savings issues facing women, African-Americans and Latinos. In addition, I will present both some general solutions and some specific recommendations to help resolve these issues.
There are similarities in the reasons that all three groups under save, but it would be a serious mistake to assume that one solution will solve all problems. In addition to specific economic factors such a job type or tenure, there are also cultural causes of each group's behavior that must be addressed to improve its retirement security. However, the cultural and economic situations differ from group to group, and even within groups. For instance, it is a mistake to assume that all Latinos fit the same behavior pattern and economic circumstances despite the fact that many are born here, while those who are not come from a number of specific countries. Clearly, the same is true for both African Americans and women.
For that reason, consideration of solutions should start with mass approaches such as automatic enrollment and similar features, which I will discuss in a moment, but then turn to specific educational outreach, mentoring, and financial literacy efforts aimed at specific populations. I strongly believe that automatic enrollment in a DC environment is an essential step, but it may not be sufficient to improve retirement security. Instead, it should be supplemented with additional efforts.
Let me also be clear that while most of my testimony will focus on the DC system, since it is the predominant retirement benefit offered by employers, I am by no means saying that it is perfect. For some populations, a DB plan may have distinct advantages, but to date there is no sign that the trend away from such plans is lessening.
Finally, while many discussions focus on the accumulation stage since no good outcome is possible in a DC plan without sufficient savings equal attention must be paid to what happens when an individual actually nears retirement. If this country is to continue to have a DC system and fails to increase the use of guaranteed lifetime income products, it runs the real risk of having hundreds of thousands of older retirees running out of money. This issue must receive equal weight in future reform efforts.
Retirement Savings Experience among Women
Women have experienced substantial gains in the labor market over the last several decades. The share of women in the labor force has grown from under 38 percent in 1960 to almost 60 percent in 2000. Women have also made concomitant gains in educational attainment levels and wage rates. Today, a higher proportion of women than men graduate from college and women's earnings are approaching the level of men's. These gains made by women have been driven, in part, by institutional changes that created employment opportunities for women and, in part, by changes in social norms that transformed the perception of women's work from a "job" to "career" and galvanized women's participation in the labor force.
Despite the improvements in women's employment outcomes, gender differences in employment persist in several key aspects. First, women are more likely to choose jobs that are part-time, have shorter careers in the paid job market, and experience shorter job tenure at any given point in time than men. Second, despite the fact that many women have entered highly-skilled and highly-paid occupations the majority of women still work in occupations or industries with lower earnings. Women continue to account for a higher proportion of workers in service and sales/office occupations, which tend to have lower earnings relative to other occupations. Even among professional workers, women are more likely to be employed in professions with lower relative earnings, such as education, training, and library occupations, rather than computer and mathematical occupations.
These gender differences in employment patterns partly explain women's lower earnings relative to men. Women's wages remain 20 percent lower than men even among full-time workers with comparable education attainment and age. Between 1979 and 2005, the difference between men and women's hourly wages (gender wage gap) shrank by almost half. Most of the decline occurred during the 1980s. The shrinking wage gap in the 1980s is largely attributable to women's increasing labor force attachment and market skills (from education and experience).
These gender differences in employment and wages lead to lower overall retirement saving for women compared to men. The last 30 years has seen a shift in employer provided retirement coverage from DB to DC plans. In a DC plan, which emphasizes accumulating assets, women are able to save less than men because they have shorter careers and lower wages.
Comparing retirement accounts of women and men, women near retirement are 5 percentage points less likely than men to have a pension or a retirement plan (such as a 401(k) and IRA). Women also have lower retirement assets than their male counterparts: the median female worker near retirement held $34,000 in a 401(k) plan or IRA whereas her male counterpart held $70,000.
Accounting for differences in employment patterns, removes much of the gender difference in men's and women's saving pattern. Not only do women have comparable participation and contribution rates to men, at each earnings level, female wage and salary workers are slightly more likely to participate in a pension or retirement plan than male workers and the difference is largest among workers in the middle- and lower income ranges. For instance, in 2005, 58.2 percent of female wage and salary workers participated in an employer plan compared to 55.4 percent of male wage and salary workers and among employed workers ages 18-62, women contributed 7.2 of their salary to a DC retirement plan while men contributed 7.5 percent.
Differences in retirement balance may also be due to differences between men and women in investment patterns. Participation in 401(k) plans requires management of investment accounts. If women are more likely to invest in less risky assets than men, they will experience lower returns on their 401(k) investments, which lead to lower 401(k) balances over time.
Although some studies have found gender differences in risk-taking behavior, the evidence is mixed and inconclusive. The shift away from DB plans to DC plans has affected more than just women's retirement balance sheets, and these other changes have both helped and hurt prospects for women in retirement.
Compared to DB plans, 401(k) plans offer greater portability, faster vesting, and faster accrual of benefits, all of which are better suited to women's interrupted work history and shorter job tenure. Pension benefits in a DB plan typically increase with earnings and years of service with a firm. As a result, they penalize those with short job tenure, since benefits at a particular job accrue at rates that are proportional to job tenure and since benefits "start over" in a new job. In addition, DB benefits vest more slowly than 401(k) balances. In a 401(k) plan, employees' contributions are vested immediately and employers' contributions under DC plans tend to be vested earlier than under DB plans.
The major disadvantage for women of the shift away from defined benefit plans and toward 401(k) plans is the loss of the automatic life annuity through an employer-based retirement plan. DB plans must offer (as a default) the option of benefits in the form of a life annuity, and often pay benefits in that form. In contrast, 401(k) plans generally provide a lump-sum distribution at retirement (in 2005, only 20 percent of employers with 401(k) plans offered an annuity payout option). Because women tend to live longer than men, a life annuity, which insures against outliving one's resources, is more valuable to women than to men.
Although one could use the lump-sum distribution to purchase a private annuity, markets for individual annuities are poorly developed and feature high expenses, making such investments unattractive. Private annuity contracts are a particularly bad deal for women because they have longer life-spans than men and, consequently, face relatively higher prices for an annuity that pays a fixed amount per year for life. This type of disparity does not exist under a DB system where men and women would receive similar benefits over their lifetime if they have similar employment histories.
An additional disadvantage of DC plans for women is that generally spousal consent is not required when the retired worker makes distribution choices at the distribution date. Under traditional DB pension plans, benefits to married workers are automatically paid as a lifetime annuity with survivor benefits for the spouse unless the spouse consents to waive the survivor benefits. By contrast, under a DC plan, there is no default distribution option and a worker may choose to take distributions as a lump sum or in installments without the spouse's consent. Men and women, however, will likely have different preferences regarding the form of the distribution because of differences in the length of their retirement period.
Requiring spousal consent when the worker makes distribution choices potentially could increase the proportion of workers taking distributions in the form of a life annuity with survivor protection. Evidence indicates that when the default option in DB plans for married couples was changed to a joint and 1/2 survivor annuity, unless the spouse consented to an alternative option, the selection of survivor annuities by married male pension plan participants increased from 48 to 64 percent.
On the other hand, 401(k) plans have become increasingly electronic, which has the potential to reduce administrative costs. Spousal consent proposals, by calling for a spouse's signature that is notarized or witnessed by a plan representative, generally have been viewed as precluding electronic administration in this phase of 401(k) plan operations. Accordingly, plan sponsor representatives have expressed concerns that expanding 401(k) plan spousal consent requirements could increase administrative complexity and costs. This issue has been the subject of considerable discussion and controversy for years. It would be useful to continue this discussion and explore approaches that could balance the legitimate interests in protecting spouses, promoting lifetime guaranteed income and minimizing 401(k) costs and administrative requirements.
Marriage patterns and living arrangements have changed considerable over the last half century. Fewer adults are married, more are choosing to divorce or remain single, or live in cohabiting households. Marriage rates have fallen from 77 per 1,000 unmarried women in 1970 to 41 in 2005. In recent years, most of the decline in marriage rates has occurred among households with lower educational attainment. The rise in single motherhood is also notable. The percent of all births to unmarried women has increased dramatically, rising from 5 percent in 1960 to 37 percent in 2005.
Marital patterns vary by race. Among white women, the percent currently married declined from 67 percent in 1960 to 54 percent in 2006. Among African American women, the decline in the proportion currently married was even more sharp - falling by nearly half from 60 percent to 34 percent. There are also large racial differences in the percentage of non-marital births. In 2005, 69 percent of births to African American women and 48 percent of births to Latino women were outside of marriage whereas only 25 percent of births to white non-Hispanic women were outside of marriage.
The decline in marriage rates creates concerns for women's retirement security because of the close link between marital status and economic status for women. Unmarried women, on average, have fewer economic resources than married women. Near or nearly retired unmarried women are three times more likely to be poor and have lower household income and net worth than similarly-aged married couples. Even compared to unmarried men in the same age group, unmarried women are financially worse off.
Unmarried women from minority groups have even lower economic resources: nearly 30 percent of unmarried African American and Latino women are living in poverty and they have between 10-25 percent the net worth of unmarried white women. Single mothers are particularly vulnerable to living in poverty than other types of households with children. In 2006, 37 percent of female-headed households with children under the age of 18 had income below the poverty-line compared with 18 percent of male-headed households and 6 percent of married couples.
The importance of marital patterns and living arrangements for economic welfare persists into the retirement years. Elderly widows are three times as likely to be poor as elderly married couples. This is partly because widowed households are more likely to have incurred large out-of-pocket medical expenses from their husband's illness. Additionally, households in which a husband dies at a relatively young age may have lower resources even prior to widowhood than households in which both spouses survive. One study found that forty-four percent of the difference in economic status between widow(er)s and married elderly persons was due to disparities in economic status that existed prior to widowhood.
Retirement Savings Experience among African-Americans2
Many studies have documented that the average African American family has lower savings than the average U.S. family. The median net worth of African Americans is only around $28,000 compared to $140,000 for the median household nationwide. The difference in savings between black and other households is even larger if one excludes the value of housing equity. The value of non-housing assets for the median black family is one-twelfth that of the typical American family, compared to one-fifth when housing equity is counted.
Racial differences in retirement resources are also apparent among current retirees. The median African American family aged 70 and older has a net worth of $36,900, which is markedly lower than the net worth of the median U.S. household. In addition, sources of income also differ between black and white retirees.
Retired African American households are more likely to rely on income from pension plans and earnings than other retirees. Furthermore, income from assets represents a much smaller portion of total household income for African Americans than for other households (almost half as large).
To a large extent, observed racial differences in retirement resources. To a large extent, these differences are attributable to racial differences in labor market experience, such as a higher unemployment rate and more part-time work. These differences lead to lower average earnings than white persons and differential access to employer-sponsored retirement plans. Some of the differences in retirement resources, however, are also attributable to racial differences in saving decisions. For instance, 401(k) participation and contribution rates are lower among black workers than white workers, and investment choices are sometimes different. In most (although not all) instances, these differences in labor market experience and saving choices adversely affect African Americans' retirement security.
A. Differences in labor market experience
African Americans are more likely to be unemployed than white persons. Unemployment rates are 50 percent higher among African Americans than white persons. In fact, higher unemployment rates among African Americans are apparent across all education groups, including among college graduates who, as a group, tend to have the lowest unemployment rate among all workers. Even among those who are employed, labor market outcomes between African American and white workers differ. A larger fraction of black workers are employed in low-wage occupations than other workers: 15 percent of low-wage workers are African Americans, yet African Americans comprise only 11 percent of the total workforce. In addition, among men, African Americans are more likely to work part-time than whites: 13.2 percent of black men work part-time compared to 12.3 percent of white men.
As a result of these labor market differences, black families tend to have lower earnings than white families. As a group, black men earn about 72 percent of what white men earn and black women earn 86 percent what white women earn.
Low-earning households save less (in levels) than high-earning households because they have a lower base from which to save. In addition, there is some empirical evidence that suggests that households with low incomes tend to save a smaller fraction of their incomes than high-income households. Thus, the combination of a lower base for saving and lower savings rate partly explains the racial gap in net worth. Differences in earnings also have implications for retirement resources because retirement balances and pension payments often are tied to earnings.
Under a DC retirement plan, such as a 401(k) plan, the employer's contribution to the plan typically is a fixed percentage of the worker's earnings. Consequently, African American workers with lower earnings would receive lower levels of employer contributions than other workers with higher earnings. Likewise, DB plans typically compute pension payments based on earnings and years on the job. Accordingly, workers with lower earnings accrue lower pension benefits than higher earning workers and, workers with shorter tenure on the job accrue lower pension benefits than those with longer tenure. On average, African American workers have lower earnings and they have shorter job tenure than other workers; which means they are more likely to accrue lower pension benefits in DB plans than other workers.
Racial differences in labor market experience also lead to differential access to employer sponsored retirement coverage. In the private sector, black workers are less likely to be offered retirement coverage than other workers because they are more likely to work in part-time and lower-wage jobs, which typically do not offer coverage. The private-sector coverage rate for African Americans is about 3 percentage points lower than the national average.
Compared to white workers, the gap is even greater: 50 percent of white workers in the private sector are covered by an employer-sponsored retirement plan compared to only 42 percent of African American workers. Furthermore, private-sector African American workers may become increasingly less likely to have coverage relative to other private-sector workers.
Between 1979 and 2004, coverage rates in private-sector employment fell by about 10 percentage points among workers with a high school education (to about 40 percent), but stayed stable at about 60 percent for college-educated workers. Since a higher proportion of African American workers have a high school education or less, to the extent that the trend continues, the gap in coverage rates between African American workers and white workers in the private sector may increase.
Black workers, however, are more likely to be employed in the public sector than other workers. This difference in public sector employment places black workers on a more favorable footing for retirement than other workers because most public sector employers offer pension coverage and the coverage is typically in the form of a DB plan. One of the advantages of DB plans over 401(k)-like plans is that workers generally are not responsible for making participation and contribution decisions: they are automatically enrolled in the plan if they are eligible, contribution rates are pre-determined and the assets are professionally managed. Generally, these features are beneficial to all workers (because inertia can prevent or delay a worker from participating) but they are particularly beneficial to low income African Americans who, as a group, tend to have less financial market experience and lower financial literacy than other workers.
The other advantage of DB coverage over DC coverage is that benefits from a DB plan are payable for life (that is, as an annuity); whereas, the majority of DC plans do not even offer the option to annuitize plan balances. Retirees in DC plans, therefore, must manage their assets to last their lifetime and face the possibility of outliving their resources should they live longer than expected. Thus, African American retirees who, on average, have greater access to DB pensions are better protected from the risk of outliving their resources than workers with DC-type retirement coverage.
B. Differences in saving decisions
While racial differences in earnings and access to employer-sponsored retirement plans lead to racial differences in retirement account balances, the choices that black workers make regarding retirement saving may further amplify these differences. Even when black workers are offered employer-sponsored retirement coverage through 401(k)-type plans, they make choices about (i) whether to participate in 401(k) plans, (ii) how much to contribute, and (iii) how to invest their saving, that lead to lower relative savings in these retirement accounts.
For example, among workers who have access to 401(k)-type plans, black workers are less likely to participate in these plans than white workers. One study found that in 2004 only 50 percent of eligible black private sector workers participated in their employer's retirement plan compared to over 59 percent of white workers. By not participating, workers are also losing any employer matching contributions they might have received had they contributed to their 401(k) plan and the tax advantages that come with saving in these employer sponsored retirement accounts.
There also is some evidence that 401(k) contribution rates among African Americans with DC plans are smaller than those of other households.23 Estimates suggest that among workers who participate in a401(k) plan, the average 401(k) contribution rate for African Americans is 4.2 percent, while it is 5.5 percent for whites and 5.4 percent across all DC plan participants regardless of race. This same study reveals that more than half of African Americans contribute 4 percent or less of their salary to their 401(k) account.
Additionally, African Americans appear to be less likely than white workers to invest in and hold high-yielding assets, such as stocks, than the average U.S. family. To the extent that investment returns are lower as a result, the accumulated savings over time for a black worker would be lower than that of a white worker, on average. This difference in investment choices may partly explain the difference in accumulated assets between retired African Americans and comparably-aged households. At the median, African American households age 70 or older hold about $300 in financial assets compared to $13,000 held by the median retired household nationwide.
C. Income-Adjusted Differences in Retirement Resources
It is important to note that the comparisons of net worth and other retirement resources discussed in the previous section have been between groups with potentially very different income. As noted earlier, the earnings of black and white workers differ. Moreover, Black families have lower income than other families, on average, and a disproportionately higher fraction of black families are in the nation's lowest income group. The median income for African American families nationwide is about two-thirds that of the typical U.S. household ($32,000 for blacks compared to $48,000 for the typical U.S. household). In addition, 17 percent of African American families are poor compared to 7 percent of white families and the national average of 8 percent.
In order to assess the relative retirement preparedness of African American families, therefore, it is useful to juxtapose families with comparable incomes. The key question then becomes whether African American households will have comparable retirement resources as other households with similar income. In this section we examine wealth levels of households, controlling for income. This income adjusted comparison of wealth also would illuminate whether there are factors beyond income that differentially affect black and white households' retirement security. A cursory comparison of income ratios (two-thirds) to wealth ratios (ranging from one-twelfth to one-quarter) suggest that other factors may play a role.
The wealth gap persists when comparisons are restricted to households with similar income. The gap is narrower for households in the highest income group (at the median, the wealth ratio is between one-half and two-thirds) but is wider for households in the lowest income group.
The typical (or median) poor African American household has virtually no savings (zero IRA/DC balances, financial assets, and net worth); whereas, the typical poor U.S. family has accumulated a small amount of wealth (primarily from equity in their home). Among middle-income households, the typical African American household still has lower savings than other households. The net worth of a typical pre-retired middle-class black family is only about half of the net worth of a typical pre-retired middle-class household in the population at-large ($47,000 compared to $95,000 respectively), indicating that even middle class blacks will be entering retirement with fewer resources than other Americans.
Similarly, black households in the top 10 percent of income have lower net worth than comparable households. While the typical rich household in the United States has accumulated about $50,000 in retirement savings in IRA/DC accounts and about $90,000 in financial assets, the typical African American household in the nation's top income group has an IRA/DC balance of just $21,000 and financial asset holdings valued only at $13,000.33 Blacks therefore have a resource deficit even when one considers the most well-to-do Americans.
Few black households, however, have incomes that place them in the top 10 percent of all Americans. In fact, black households that are in the top 10 percent of income among other black households have substantially lower income and lower net worth than those in the 10 percent of the U.S. population ($257,500 compared to $708,000). Comparing liquid wealth, such as financial assets, the typical black household at the top of the African American income distribution has no more than one-eighth the assets of typical rich household. Accordingly, African Americans who may very well consider themselves rich relative to their peers still lag far behind many other U.S. households when entering retirement.
D. Retirement Adequacy
The analysis to this point has focused on household net worth primarily, which includes the value of home equity, financial assets, and savings in tax deferred retirement accounts, such as IRAs and 401(k)s. The measure of net worth, however, does not include expected claims to future benefits from Social Security and DB pensions. As was noted earlier, the importance of Social Security and pension benefits varies across households and, for some, these benefits can account for a substantial portion of retirement income.
Because Social Security benefits are progressive, in that benefits replace a higher portion of earnings for low income workers than for high-income workers, and because black households on average have lower income than other households, the omission of Social Security benefits from household resources would tend to overstate the differences in retirement preparedness between black and white households. Similarly, to the extent that black workers are more likely to receive DB benefits than white workers, the omission of DB payments would overstate racial differences in retirement resources.
A more appropriate comparison of relative preparedness of those not yet retired therefore would account for these expected sources of funding, in addition to accumulated assets. One common approach is to evaluate whether and to what extent a household's expected retirement resources (including all saving, Social Security and DB benefits) are sufficient to maintain their pre-retirement standard of living throughout retirement. That is to say, what percentage of the household's income before retirement can be replaced by accumulated savings and claims to benefits in retirement (the income replacement rate). It is generally thought that an income replacement rate of 75 to 80 percent would be needed to finance consumption during retirement and maintain pre-retirement standards of living.
Estimates of expected mean retirement income of African American and white families suggest that the average African American family could expect to have a replacement rate of 73 percent of preretirement income, whereas the average white family could expect to replace 83 percent of their pre-retirement income, when pension wealth and social security are included. Thus, even after accounting for expected Social Security and DB benefits, a gap still exists in the degree to which African Americans and white families are prepared for retirement: African Americans lie near the lower end of the range deemed adequate to maintain their living standards while whites are slightly above the top of the range.
Moreover, a recent study that examines families below the adequate range reveals that 40 percent of African Americans could expect to have a replacement rate that is lower than 50 percent; whereas only about one-quarter of white households had estimated replacement rates in that range. Thus, a significant minority of households will not be financially prepared for retirement and a disproportionate number of those households are expected to be African Americans.
E. Other Explanations for the Income-Adjusted Racial Wealth Gap
It is common to view household saving as a process governed solely by factors affecting the nuclear household however, there is increasing evidence to suggest that relationships with extended family members may affect the way that individuals prepare for retirement. Sharing resources, such as housing, may help older households meet some of their retirement needs and help some families accumulate wealth. At the same time, intra-family transfers to support poorer kin members also may depress younger generations' ability to save for their own retirement. Furthermore, knowledge and skills about saving could be shared within the family, so young African Americans from lower-income families may not have the same learning opportunities as those from higher income families.
Economists have argued that families can do almost as well as insurance markets in protecting against adverse events if family members pool risks and resources. Choosing to pool risks and resources within the family can allow an individual who suffers a loss of income to rely on other family members to help finance consumption when he or she experiences the loss in income, which mitigates the need to save and shore up contingency reserves for the situation.
It is well-documented that minority families, including African American families, are more likely to live in extended households than non-Hispanic white families. Nearly one in four African Americans lives in an extended household compared with only one in ten non-Hispanic white persons. Furthermore, these racial differences in extended family living arrangements remain even after controlling for economic characteristics. One study finds that blacks and Hispanic households are 14 percent more likely than non-Hispanic white households to be extended.
To the extent that families that live together are more likely to help each other financially, the higher presence of extended households among African American families than non-Hispanic white families may explain part of the racial difference in wealth among those who are near retirement. Potentially, individuals may decide they do not need to save large sums independently because they believe they can rely on other family members for assistance.
Evidence suggests that among single elderly women, co-residence with other family members appears to alleviate poverty for nearly one out of every seven household, for example. Similarly, single women benefit from co-residing with other family members - regardless of their racial background. However, because elderly African American women are 1.5 times more likely to be single than elderly non-Hispanic white women and a greater proportion of younger single mothers are African American than white, extended kin networks may represent a more important resource for both young and old black women than white women.
Even individuals or families that do not reside together may choose to share resources and to pool risks. Some studies suggest that African Americans are highly embedded in kin networks. If individuals who are accustomed to turning to relatives in times of financial need can expect to rely on such family assistance during retirement, they would have less incentive to save independently for this life stage.
Financial transfers between family members
While family structure and participation in kin networks can provide a barrier against poverty and act as a kind of retirement resource for the old, other studies suggest that African Americans' entrenchment in family networks actually may depress their ability to save while they are young.
In particular, two studies find that middle class African American adults are more likely to have poor relatives than other middle class individuals, and that the presence of poor relatives throughout the family tree is negatively correlated with an individual's wealth. The authors argue that the total amount of savings that a middle class African American household is able to accumulate during its working years is reduced because of the need to assist poor parents through financial transfers.
Financial transfers from older family members to younger members, on the other hand, can be beneficial for building wealth. In the empirical research that examines the effect of bequests and inter-vivos transfers on wealth accumulation there is general agreement that the receipt of financial transfers can positively affect the recipient's ability to accumulate wealth. Financial assistance to pay for schooling, for example, will improve an adult child's future earning opportunities; and a down payment gift to facilitate the purchase of a home will reduce the amount of debt that one's offspring has to take on to purchase a home. Thus, transfers can help younger generation family members build wealth earlier than they otherwise would. The literature has shown that African Americans are less likely to receive these types of transfers, even after controlling for the economic resources of the adult child.
Most studies agree that this implies that young black families are at a disadvantage relative to other families that do receive financial assistance (at least in theory). There is less agreement, however, about the volume of transfers that occur, and whether they have large effects on children's saving behavior, and the exact extent to which they create a sizeable gap between the wealth held by African American and other households.
Another possible consequence of African Americans' greater likelihood of having poor relatives throughout the family tree is that blacks are subsequently less likely than those with wealthier kin networks to acquire knowledge, skills and opportunities related to saving and perhaps more basic understanding of financial matters. Research suggests that attitudes about saving, risk preferences and investment choices may be learned or mimicked from parents.
These wealth-enhancing "traits" are more likely to be transferred to children by wealthy parents, who have greater understanding, access to, and knowledge about financial markets, than low-wealth parents. There is also increasing evidence that financial illiteracy is widespread among the U.S. population but it is particularly acute among specific demographic groups, such as those with low education, women, African-Americans, and Hispanics.
Financial literacy can have a significant effect on retirement outcomes. Studies show that those with greater financial literacy are more likely to plan for retirement and planners are more likely to approach retirement with higher wealth levels than non-planners. Evidence also suggests that those who have low financial literacy are significantly less likely to invest in stocks. Thus, differences in financial literacy and financial market experience between African Americans and whites could explain part of the wealth gap.
What is less clear, however, is whether these differences in financial literacy and financial market experience remain after controlling for income differences. In any event, because middle class African Americans are more likely to have poor parents than white middle class families and to the extent that poor parents are less likely to be financially literate, middle-class African Americans may be less financially literate than middle-class white families.
Retirement Savings Experience among Latinos
Latinos face particular challenges in preparing for retirement. Only about a quarter of Hispanic workers participated in an employer-provided pension plan in 2001, compared to about half of the overall workforce.4 This low level of pension participation represents a threat to Latino retirement security.
The lower rate of Hispanic pension participation persists even within earnings, age, and firm size categories. For example, pension participation in 2003 was significantly lower for Hispanic workers than for White workers within any earnings category. (Note that native-born Hispanics had participation rates that were significantly higher than nonnative-born Hispanics. Participation rates for native-born Hispanics were only somewhat lower than those for White workers within the same earnings category.) Similar patterns emerge within age categories and within firm size categories.
Data on accumulated assets in 401(k)s and IRAs also point to lower retirement savings among Hispanics, including within any given income category. The 2001 Survey of Consumer Finances shows the average 401(k) and IRA balance and the median balance for all households and for Hispanic-headed households. Among all households, the median balance held in these types of retirement accounts was $600; the average was $53,670. Among Hispanics, the median was zero and the average was $10,480.
Asset balances for Hispanics are significantly lower in any given income category: among households with incomes between $50,000 and $75,000, for example, the average balance for all households was more than $50,000; among Hispanics, the average was only a little more than $12,000. The very modest account balances held by Latinos underscores the fundamental challenge of boosting retirement savings specifically for Latinos. Even when Latinos are participating in retirement savings vehicles, they do not take advantage of these options to the extent that they could. Making saving easier could increase retirement security for the Latino community.
Part of the explanation for the sharp differences is the lower rate of participation in 401(k)s and IRAs among Latinos. Even among those with accounts, however, a significant difference generally remains. In other words, Hispanics are less likely to participate in 401(k)s and IRAs, and those Latinos who do participate typically contribute less than other participants. The accumulated account balances for all households and for Hispanics when the analysis is restricted to those with an account. Hispanic households with a 401(k) or IRA tend to have significantly lower accumulated balances than all households with an account. For example, between $50,000 and $75,000 in income, the average balance for all households was more than $58,000; the average for Hispanics was under $23,000.
Finally, part of the explanation for the differences may reflect the age distribution of Hispanics compared to the overall population. Latinos are a younger population comparatively. But even among those aged 55 to 59, and therefore on the verge of retirement, Latino account balances are significantly lower than for other households. Among all households in this age range, the median combined 401(k)-IRA balance was $10,400; among Latino households in the same age range, the median was zero. In other words, the majority of Hispanic households aged 55 to 59 and therefore on the verge of retirement had nothing accumulated in either a 401(k) or an IRA.
Similarly, among all households in this near-retirement stage, the average combined 401(k)-IRA balance was almost $120,000; among Hispanics, it was just over $35,000. This low level of pension accumulation means that Social Security benefits dominate as a source of income for retired Latinos. According to the Pew Hispanic Center, 76 percent of elderly Hispanics who receive Social Security benefits rely on those benefits for the majority of their income. Perhaps even more astonishingly, Social Security benefits represent the only source of income for two in five (41 percent) of elderly Hispanic beneficiaries. This is twice the percentage for all elderly beneficiaries, where 20 percent rely exclusively on Social Security - still too high a share, but much lower than among Latino beneficiaries.
Rigorous economic analysis also suggests disproportionate under-saving among Hispanics. For example, Engen, Gale, and Uccello incorporate the implications of uncertain wages into their analysis of retirement savings adequacy. The analysis recognizes that because a household's future income is uncertain, the level of current assets necessary to live comfortably in retirement is also uncertain. They therefore generate a distribution of optimal wealth targets relative to earnings for narrow classifications of households (separated by age, education, pension status, marital status, and current wage). They then compare actual wealth-earnings ratios to the simulated optimal targets, and see what share of households are above the median simulated optimal target. If every household were saving the right amount for retirement, and the Engen-Gale- Uccello model were exactly correct, half of households should have wealth earnings ratios in excess of the median simulated optimal ratio for their household type. To undertake these comparisons, the authors apply three definitions of "wealth." "Broad wealth" is equal to all net worth other than equity in vehicles. "Narrow wealth" is broad wealth excluding equity in the household's primary residence. "Intermediate wealth" is broad wealth excluding one-half of the household's equity in its primary residence.
The Engen-Gale-Uccello results using the 2001 Survey of Consumer Finances for all households and for Latino households shows, much smaller percentages of Latino households than other households are at or above their median simulated optimal wealth-earnings ratio. For example, using the narrow wealth measure, 52 percent of all households are at or above the median simulated wealth earnings ratio for their household type. Among Hispanic households, however, under 20 percent are at or above the median. In other words, under this more rigorous analysis, as under the simple asset calculations above, Latinos appear to be disproportionately saving inadequately for retirement.
According to official projections, Hispanics have higher life expectancy than other Americans. At age 22, for example, Hispanics have a life expectancy of 60 years - three years longer than White non-Hispanics and more than seven years longer than Black non-Hispanics. At age 65, Hispanics have a life expectancy of more than 21 years, again significantly longer than non- Hispanics. Such relatively long life expectancies for Latinos reinforce the concerns about their savings adequacy: longer life expectancies, unless offset by later retirements, increase the accumulated assets needed in order to live comfortably throughout retirement.
Studies like that of Engen-Gale-Uccello assume that Latinos have the same life expectancy as other people; to the extent that Latinos actually have longer life expectancies, their retirement saving adequacy is even worse.
In evaluating the official life expectancy figures, it should be noted that researchers have raised questions about the longer-than-average life expectancies among Latinos. In particular, life expectancy for native-born Hispanics appears to be similar to that of non-Hispanics. The differences appear to arise solely from nonnative-born Hispanics, and it is possible that the life expectancy differential for non-native-born Hispanics reflects measurement errors.
The bottom line is that too many Latino families are failing to save adequately for retirement.
Strategies to Improve Retirement Savings
A. Financial Mentoring
Research suggests that attitudes about saving and investment choices may be learned from parents or mimicked and, thus, would be important for the accumulation of savings - but children from moderate- and low income families often lack access to this source of information. This suggests that one possible strategy to increase retirement saving and investment outcomes would be to design policies that remedy these information gaps. Then, individuals who are born into homes with limited information would be able to save as effectively as those born into families with financial market expertise.
Attitudes about saving are likely formed from an early age. Habits learned from when one is young tend to carry into adulthood. If children are encouraged at a young age to start saving and notions of budgeting are integrated into their daily lives, it may instill a positive attitude towards saving over their adult years. Furthermore, if children are introduced to financial institutions and financial terminology and given an opportunity to make financial decisions, it may help overcome barriers (either informational or psychological) that have previously inhibited participation in financial markets.
One approach, therefore, might be to develop a mentoring program targeted at very young children. The approach might couple young children with a financial institution that provides ongoing financial mentoring to the child over a period of time. The mentoring would begin at a very young age - as early as age 6 - and continue until the child reaches age 18. At very early ages, children would be given simple financial goals, such as opening a savings account. As they get older and become more familiar and comfortable with money management, the mentoring would include discussions about saving with a target or for a particular outcome, long-term versus short-term saving goals, access to credit and paying down debt, and, if account balances are large enough (for older children), diversifying funds. It also could include discussion of realistic expectations about retirement and sources of retirement income.
B. Automatic Saving
Financial mentoring alone, however, would be insufficient to ensure retirement security for many Americans. Many individuals, even those who are financially savvy, are at times too busy to take the time or the effort to sort through complicated financial decisions. As a result, they either postpone making these decisions or rely on simple rules of thumb that sometimes do not result in prudent investment choices.
For instance, many workers are offered the opportunity to save through their employer's 401(k) plan, yet many do not participate in the plan. In many 401(k) plans, workers must make an active decision to participate in the plan. However, participating involves more than simply filling out a form (either paper or electronic). Workers must decide how much to contribute and which investment option provided by their employer to select. These decisions are complicated and make the 401(k) enrollment decision daunting, particularly for workers who have limited investment experience. As such, many workers postpone or avoid making the enrollment decision.
Non-participation is particularly high among lower-income and minority workers. It is also higher for African Americans than white workers. By not participating in the 401(k) plan, workers miss the opportunity to benefit from any employer matching contributions and the tax savings from contributing to a 401(k). Furthermore, by delaying participation, they lose the opportunity of letting compounding returns grow their assets over time.
A strategy that has proven successful at getting workers to participate in their 401(k) plan is to change the default option from an opt-in system to an opt-out system. Under an opt-out system, workers are automatically enrolled into their company's 401(k) plan, unless the worker affirmatively chooses not to participate. With automatic enrollment, the employer selects a default contribution level and at each pay-period, that portion of the worker's paycheck is directed into a default investment fund.
Automatic enrollment in 401(k) plans has been shown to increase participation significantly: in firms that implemented automatic enrollment, participation rates increased from 75 percent to as high as 95 percent among new hires. The effect was even more dramatic among women, minority and low-income workers - increasing participation rates from 35 percent to 86 percent for women and from 13 percent to 80 percent for workers earning under $20,000 a year.
With automatic enrollment in 401(k)s, workers who are unable or unwilling to make decisions about their 401(k) participation (because it is too complicated or they do not have the time) would still be able to participate in the plan. Automatic enrollment harnesses the power of inertia so that individuals save even if they make no effort.
Automatic enrollment in 401(k) plans is most effective when it is combined with other automatic features. These include automatic enrollment with automatic increases in contribution rates over time, automatic investment in prudent and diversified funds, and automatic rollover of 401(k) assets when the worker leaves a job. Plans with these enhanced automatic features ("second generation" plans) would help workers gradually save more over time, improve investment returns in their portfolios and retain assets in the retirement system when they change jobs.
Many "first generation" plans typically enroll only new workers at a 3 percent contribution rate and do not increase contribution rates over time. The typical default 3 percent rate, however, is less than half the average pre-tax contribution rate of 7 percent of pay among participants in non-automatic 401(k) plans. Furthermore, the majority of participants who are enrolled under automatic enrollment tend to remain at the initial contribution rate if the company does not automatically increase their contribution rate over time. Inertia keeps workers from saving more over time.
Building in annual increases in contribution rates (automatic escalation) would help workers gradually save more over time with virtually no effort. Contribution rates could be increased by one percent per year, unless the participant opted-out of those increases.
Additionally, a "second generation" plan might enroll new and current nonparticipating workers at a higher initial contribution rate, such as 5 or 6 percent. These enhancements would help workers save more and take advantage of any additional employer matching contributions which they may have previously missed. Currently, about one quarter of employers that offer automatic enrollment also offer automatic escalation. Wider adoption of automatic escalation would, over time, improve the financial position of many lower-income families, who tend to have lower saving rates than higher income households
Under automatic rollover, when an employee changes jobs, the funds in her 401(k)-type account would be automatically rolled over into an IRA or another retirement plan offered by the new employer. The worker, however, can affirmatively choose not to roll over his or her funds. Many workers, particularly those with small account balances, tend to cash out their accounts when they leave a job; thereby, losing the value of the tax-preferred saving and leaving them with less retirement saving.
Empirical evidence suggests that simply changing the frame within which workers evaluate their options for preretirement distributions can retain assets in a tax-preferred retirement accounts and prevent "leakage" from the retirement system. In this way, automatic rollovers help to keep retirement funds in the retirement system when workers change jobs by making rollovers the default option.
About half of all workers do not have access to employer-sponsored retirement plans because they are employed by companies, usually small businesses that do not offer any type of pension or retirement savings plan. These workers would not benefit from the automatic 401(k).
However, a bipartisan, cross-ideological would create Automatic IRAs, a system whereby workers without access to employer-sponsored retirement plans can contribute automatically to a low cost, diversified IRA through direct payroll deposits. Because IRAs are portable and are not tied to a particular employer, employees would be able to continue contributing to the IRAs even when they switch jobs. Under this proposal, workers would be enrolled in an IRA and deposits to the IRA will be made automatically at each pay period, unless the employee actively chooses not to participate in the program.
A firm that is not ready to adopt a 401(k) or other retirement plan would offer its employees the ability to save in an IRA every payday by payroll deposit, which is currently available to millions of employees. Once payroll deposits begin, they continue automatically unless the worker later opts out. Employers above a certain size (e.g., 10 employees) that have been in business for at least two years and do not sponsor any retirement plan for their employees would be required to offer employees this payroll-deduction saving opportunity.
The Automatic IRA would involve no contributions or other outlays by employers. Employers would merely offer their payroll system as a conduit that employees could use to save part of their own wages in an IRA. Participating employers would receive temporary tax credits, would be required to obtain a written waiver from any employee who does not participate, would be encouraged to use automatic enrollment and would be able to protect themselves from fiduciary liability. Employers could designate the IRA to receive the savings, including, as a fallback for those unable or unwilling to choose, a national platform IRA where the account would go to a private sector provider that has agreed to take accounts from any company.
The default investment for the Automatic IRA would be a diversified, low-cost life cycle fund, with other choices available. The self-employed would be encouraged to save by extending payroll deposit to independent contractors, facilitating direct deposit of income tax refunds, and expanding access to automatic debit arrangements linked to IRAs, including on-line and traditional means of access through professional and trade associations.
C. Saver's Credit
While 401(k)s and IRAs represent effective instruments for saving for retirement because of their tax-preferred status, for many low income families, the incentives to save in a tax preferred retirement account may be small. Contributions to tax-preferred retirement accounts reduce the saver's taxable income; thus, they generally benefit those with higher marginal tax rates more than those with lower marginal tax rates. Low income families tend to face low marginal tax rates or to pay no taxes at all. They therefore stand to benefit less than other workers from tax incentives that reduce taxable income, or not at all.
Tax credits, however, can be used to create strong incentives for low- and moderate income households. For example, the Saver's Credit, which specifically targets moderate- and lower-income families, was enacted in 2001 to give taxpayers earning less than $52,000 a tax credit for contributions to 401(k) plans, IRAs, and similar retirement saving vehicles. Depending on the taxpayer's income, households can receive a credit of 10, 20, or 50 percent of their contributions to a retirement account.
The Saver's Credit, however, is currently nonrefundable: it merely offsets a taxpayer's tax liability. Therefore, it provides no saving incentive for almost 50 million lower-income households that have no income tax liability. Making the Saver's Credit refundable would provide an important incentive to these households to save, which would improve retirement security for those families with the lowest incomes. This might have the added benefit of making them less dependent on Social Security income and means tested government programs during their retirement years. There is also some evidence that restructuring the credit as a matching contribution that is automatically deposited into an IRA could further increase the incentive to save.
The Saver's Credit also could be simplified with a single 50 percent credit rate, phased out smoothly above the income eligibility limit. It could be improved to reach a larger group of families by expanding the eligibility limit to include households with income of up to $70,000 per year. This would increase the incentive to save and help moderate and lower-income Americans to save for a secure retirement.
D. Updating Asset Test Rules
One final way that policymakers can introduce reforms that improve the incentive to save is by modifying the rules for eligibility for public assistance program. Outdated asset tests in means-tested public assistance programs (such as Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF) and Medicaid) penalize lower- and moderate-income households that save. There programs help individuals in times of unexpected hardships or illness, or who may not have had a full working history and rely on these sources of public assistance to supplement their retirement needs.
To be eligible for these programs, applicants generally must meet an asset test as well as an income test. While the asset tests usually do not count accrued benefits under a defined benefit plan as assets, many states often count 401(k) or IRA balances or both. This has the effect of a steep implicit tax on 401(k) and IRA saving. As a result, pre-retired families with incomes low enough to qualify for a means-tested program under the income test might respond by saving less. Similarly, families that suddenly experience a temporary disruption in employment would be required to nearly deplete their retirement savings before becoming eligible for the public assistance.
Although some state programs have eliminated asset tests, or at least aligned the treatment of defined contribution plans with that of defined benefit plans, many have not. The 2008 Farm Bill addressed the inconsistent treatment of assets in tax preferred retirement accounts in the food stamp program by excluding those assets from the asset tests. However, asset tests in other public programs, such as SSI and Medicaid, continue to treat retirement saving in a confusing and seemingly arbitrary manner, with different restrictions state-by-state and account-by-account.
It would be far better policy to eliminate the implicit tax on retirement saving in these programs by mandating that retirement accounts such as 401(k)s and IRAs be disregarded for eligibility and benefit determinations in federal and state means-tested programs. Changing the law to exempt retirement accounts from being considered in means-tested programs would treat retirement savings fairly and consistently and would send an important signal to families that rely or might need to rely on means-tested programs in the future: you will not be penalized for saving for retirement.