6

Retirement Benefits

It is widely discussed in the financial pages of media outlets and financial advice columns that Americans do not save enough for retirement. A common view of what constitutes “enough” is a nest egg sufficient to allow a person to roughly maintain their standard of living after retirement—or at least avoid a drastic reduction. But meeting even modest retirement goals requires planning. To maintain even a roughly equal standard of living during retirement, a family with an income over $25,000 should be saving about 13 percent of their income (factoring in expected social security benefits); the challenge is stiffer for higher-income families (Bernheim et al., 2000).

Average savings rates for U.S. households are far below this level: the average is 7–8 percent. According to the Federal Reserve Bank, approximately one-fourth of adults who have not yet retired have no retirement savings, and just over half of nonretired adults assess their savings as being on track (Federal Reserve, 2022). White adults are more likely than Black or Hispanic adults to have some retirement savings and to regard their savings as on track. While older adults have the greatest savings, the group with the most savings (ages 65–74) had only a median of $164,000 in retirement savings in 2019, and people ages 45–54 had only a median of $100,000 (Federal Reserve, 2021). It has been estimated that only about half of U.S. households can expect to sustain their standard of living after retirement (Osterland, 2022).

The failure to save adequately for retirement has been ascribed to many of the behavioral factors discussed in Chapter 3, particularly limited attention, cognitive barriers, present bias, and inaccurate beliefs. Limited

attention and cognitive barriers lead people to devote inadequate time to important financial decisions and limit their understanding of the ramifications of their choices. Present bias leads people to give insufficient weight to the future implications of current decisions about financial matters. Low levels of financial literacy in the general population exacerbate the effects of cognitive barriers and limited attention, and lead to inaccurate beliefs about the importance of saving (Lusardi & Mitchell, 2014).

Employers play a critical role in most people’s retirement planning, and over the last several decades most U.S. employers who offer retirement benefits have moved from defined benefit plans to defined contribution plans. Under a defined benefit plan, a company or firm combines funds deducted from employee paychecks with company-contributed funds in an account that can grow and eventually pay a specific dollar amount per month or year to retired employees. In general, such plans can yield a payout roughly equal to the individual’s salary, if the company-funded portion is sufficient. These plans have tax advantages both to workers and their employers and can provide workers with significant nest eggs. A worker with a salary of $50,000 who devotes six percent of their salary to a plan saves $3,000 in post-tax income and therefore saves that amount times their marginal tax rate (e.g., a 10 percent gain if they are in a 10 percent tax bracket). More important, if their employer matches that contribution at, say, a 50 percent rate, then the $3,000 saving turns into a $4,500 saving, equivalent to a 50 percent rate of return on top of the 10 percent tax gain. These are enormous rates of return for any form of savings.

However, the large majority of companies today have shifted to defined contribution plans, in which workers decide how much will be deducted from their paychecks, and the firm matches that deduction, according to a defined ratio. These plans are optional, and workers are not required to join them. In traditional economic models of financial decision making, it is assumed that people who are offered alternative retirement plans by their employers will rationally evaluate the financial advantages of those alternatives. That is, they will consider how much they want to contribute to savings plans (sacrificing current spending power), as well as the after-tax rate of return on any savings plans they are offered (which determines how much they will have to spend later). In making these calculations, workers would ideally take into account their particular retirement needs, the riskiness of their future income profiles, uncertain life expectancy, and other factors. This assumption has been shown to have limitations.

AUTOMATIC ENROLLMENT

Research conducted in the last two decades has demonstrated that behavioral factors not accounted for in the traditional economic model

strongly affect people’s decisions about enrolling in pension plans. A 2001 study identified striking contradictions to the traditional economic model and was followed by a significant body of work that eventually influenced government policy regarding retirement savings (Madrian & Shea, 2001). The authors studied choices by employees who were offered the option to enroll in an advantageous company 401(k) plan that guaranteed a good rate of return, especially when tax savings were included in the calculation. The offered plan also included a company match for each dollar the employees contributed (up to a maximum), which strengthened the already strong case for investing in the plan. The company required employees to actively indicate that they wished to enroll in this attractive plan and to make payroll contributions to it: the “default option”—that is, the option that would apply if the worker took no action—was to not enroll. The company examined in the study then changed the architecture of the arrangement by making enrollment in the plan the default option—that is, employees had to actively indicate that they did not wish to enroll if that was their decision.

According to traditional economic assumptions, the difference in what is set as the default—to enroll in the new plan or not—should have no impact, or at least not more than a small impact, on the decisions of rational people who have carefully considered both plans. However, the authors found that making automatic enrollment the default dramatically increased employees’ enrollment in the plan, from 49 percent to 86 percent. The default option in this case also designated a specific amount for payroll deductions—usually a small percentage, three percent, of wages—but employees could choose a lower or higher amount if they wished. The authors found that many chose the default level of the contribution, which often led to increased savings from that source as well. This was a clear sign that employees were devoting limited attention to their pension plan arrangements.

Numerous subsequent studies strongly confirmed these findings, showing major increases in participation in 401(k) plans (the most common defined contribution plans that employers offer) when the default was to enroll rather than to not enroll. These findings held across a wide variety of companies, types of plans offered, company matching policies, and other variables (Beshears et al., 2009, 2018). Indeed, an important aspect of this finding is the extent to which it has been replicated in different contexts, which confirms its validity. In addition, like many behaviorally inspired interventions, default approaches are quite inexpensive to implement, making them a particularly attractive policy option.

This research attracted significant public attention that led to provisions in the landmark Pension Protection Act of 2006, which formally allowed employers to change the default in their tax-preferred pension plans, codified the rules regarding the required characteristics of default plans to qualify for favorable tax treatment, provided employer protection

from losses from the plans, and generally encouraged employers to adopt automatic default plans (Beshears et al., 2010). Legislation passed in 2019 allowed employers to set higher maximum default contribution rates, which research has shown has a powerful effect on levels of savings. Additional legislation in December 2022 extended automatic enrollment to a wider set of employers; relaxed the penalties for one-time emergency withdrawals; and, in a provision that particularly highlights behavioral economics, allowed employers to offer workers a “small gift” to enroll in their company’s plan. Automatic enrollment has been described as “the most successful contribution of so-called behavioral economics to public policy” (Porter, 2016).

STRATEGIES TO REFINE AUTOMATIC ENROLLMENT

While automatic enrollment in tax-advantaged retirement plans has had demonstrable benefits for workers, a number of questions remain. One is about the size of contributions employees make to the company retirement plan, if they enroll in it, and how the money employees contribute is invested, which is also subject to employee choice. In the initial Madrian and Shea study (2001), the company’s default plan was for employees to contribute three percent of their earnings to the plan and specified that those investments be invested in a money market fund. Yet, as noted above, 13 percent or more is needed to maintain a preretirement standard of living in retirement, and money market investments are well-known to have much lower long-run rates of return than investments in stocks and bonds. It is even possible that people who were presented with the traditional plan (in which enrollment had to be actively selected) might have chosen contribution rates higher than three percent, and investments in higher-return investment instruments, than those offered in the company’s default plan.

One way to address this problem is simply to increase the percentage contribution in an employer’s default plan. A study of the effects of raising the default contribution percentage from three percent to six percent showed that it increased saving and did not significantly reduce plan participation (Beshears et al., 2009). Another way that firms have addressed this issue has been to offer a three-way choice instead of a two-way choice: (1) the default option of enrollment with a fixed contribution percentage and investment allocation, (2) nonenrollment, or (3) enrollment with a different a contribution percentage and investment allocation (often called “personalized enrollment”). Offering this third alternative alongside the other two makes explicit the idea that employees may wish to actively think about their contribution percentage and their investment allocation: that is, it attempts to make them pay more attention to it and encourages them to do so.

To address the possible concern that employees may be reluctant to immediately jump to a relatively high contribution rate, an alternative

known as autoescalation has been developed. In such a plan, contribution rates start off relatively low but rise automatically over time. A behaviorally inspired default plan that uses this approach is the Save More Tomorrow plan, pioneered by Benartzi & Thaler (2001). Employees that opt for this plan have their savings rate in a 401(k) plan increased by a fixed amount, say two percent each year, by default, up to a specified ceiling. This plan encourages people to save more over time without having to start saving more immediately, which is often a hurdle. Employees may opt out at any point. This type of plan addresses the behavioral trait of present bias (see Chapter 3). In its initial implementations this plan was very successful in raising savings rates of participants, and some features of this plan have been implemented nationwide.

A second issue is that people sometimes want to withdraw funds from a retirement plan for immediate consumption needs, despite the typically high penalties for early withdrawal. This is especially a concern with low-income earners who accept the default plan, with its minimum required contributions, but may later need cash to deal with temporary reductions in income or increases in expenses. This may not be a substantial problem. According to one study, only 0.3 percent of people enrolled in Vanguard plans withdrew for “hardship” reasons (i.e., the withdrawal was legally allowed but with penalties and tax due; Munnell & Webb, 2015). However, the cumulative impact for the people who made such withdrawals was an almost 25 percent reduction in the amount of money they had available when they retired. About 0.5 percent cashed out their plans when leaving their employer (this option is legally allowed), although many rolled over their balance to an individual retirement account.

Another study showed that workers with present bias often withdraw funds from their 401(k) plans when separating from their employers and also often changed their retirement assets into more liquid forms (Beshears et al., 2022b; see also Wang, Zhai, & Lynch, 2022). The researchers proposed dealing with the short-run liquidity constraint problem by establishing “rainy day” funds for employees, where automatic payroll deductions can go into a fund that is intended just for emergency withdrawals. There have also been proposals to direct employee contributions partially into more liquid buffer stock accounts, at least for initial contributions (Beshears et al., 2015b, 2020; John, 2015; Gruber, 2016; Mitchell & Lynne, 2017). Bhargava & Conell-Price (2022) have suggested that an alternative framing of such accounts as “serenity accounts” could result in greater participation.

A third issue is whether automatic enrollment plans lead employees to shift their savings from other employer plans that are less advantageous, without actually increasing the amount they are saving. This issue has not been thoroughly studied, but some evidence indicates that this may not be

a significant issue, since employees’ net savings increase, at least on average and in the aggregate. For example, one study looked at the savings effects when a group of workers accumulated sufficient years of service in their firms to become eligible for tax-favored 401(k) contributions (Gelber, 2011). The study found an increase in 401(k) balances and a statistically insignificant effect on the levels of other financial assets. This study seemed to confirm earlier work showing that savings incentives resulted in a net increase in retirement assets greater than reductions in other plans (Engen et al., 1994; Poterba, Venti, & Wise, 1995). Another study showed that automatic enrollment did not lead to an increase in borrowing, which might be expected if enrollment created financial distress (Beshears et al., 2022a).

Recently, researchers have applied other behavioral approaches to influence employees’ choices for pension plans. One study showed that even when employers offer default plans, many people do not contribute to the plans at high-enough rates to sufficiently reduce financial insecurity during retirement (Bhargava et al., 2021). The authors found that employers can increase contribution rates by using behaviorally based design features that lead employees to choose “personalized plan options”—that is, options that are tailored to an individual’s specific financial circumstances—instead of defaults. Such features include seemingly minor changes in the way the options are portrayed to employees, the language and words used, and even the color scheme of the choices. These types of changes in framing are closely related to the concept of reference dependence (see Chapter 3). Another example of personalized plan options was included in the Pension Protection Act of 2006, which allowed employers to automatically enroll employees in so-called target-date retirement funds, which make the asset allocation dependent on the individual’s age and life expectancy.

Other work has shown the benefits of framing in the presentation of alternative retirement plans to employees, offering employees personalized retirement projections, and offering financial consultation in combination with defaults (Goda, Manchester, & Sojourner, 2014; Blumenstock, Callen, & Ghani, 2018; Beshears et al., 2021). A study of the role of incentives designed to induce employees to make larger contributions and to opt for personalized plans showed that personal recommendations designed to improve financial literacy and simplified enrollment procedures had limited impacts on employees’ savings rates (Bhargava & Conell-Price, 2022). However, the authors found that small incentives designed to counter present bias were effective.

While this chapter has focused on behavioral issues in employer retirement plans, there is a larger literature on retirement savings in general that addresses the effects of behavioral interventions on savings at older ages. This work includes studies of the effects of financial education interventions on retirement savings (Fernandes, Lynch, & Netemeyer, 2014; Kaiser et al., 2022), the effects of providing peer information on individual retirement

decisions (Beshears et al., 2015a), interventions to encourage retirees not to draw down their assets too quickly (Shu & Shu, 2018), and many others. This literature shows the wide applicability of behaviorally motivated interventions to improve the financial well-being of older people.

FINDINGS

A substantial body of research has examined the role of defaults and other factors in the retirement savings decisions of workers. This research has demonstrated that behavioral factors play a key role in people’s decisions about a matter that had previously seemed to be a straightforward case for traditional economic analysis, in which logic alone would lead a rational actor to take advantage of an offered benefit. We highlight some key findings from this work.

• In making decisions about retirement savings benefits, people are strongly influenced by behavioral factors: limited attention and cognition; inaccurate beliefs; present bias; and, to a slightly lesser degree, reference dependence.
• The use of default designs by employers has been shown to have a major impact on saving for retirement, and there is little evidence of significant downsides for employees.
• Default designs have been shown to have those positive effects across a wide variety of settings, making them one of the behavioral designs with the largest generalizability.
• Other characteristics of plans, including framing, wording, and visuals, have also been shown to affect workers’ choice of plan and have the potential to at least modestly increase retirement assets.

To build on the success that has already been achieved in boosting retirement savings, it will be useful to have additional evidence on behaviorally based tools for increasing saving to levels that will better meet people’s actual retirement needs. Specifically, research is needed to address:

• ways to encourage employees to pay more attention to the exact contribution rate and form of investment in their retirement plans to allow employees to make more personalized choices rather than just accepting the default rate and form;
• methods of developing rainy day funds and other forms of liquid assets that allow workers to deal with short-term needs instead of withdrawing funds from their retirement plans; and
• ways to encourage employees who separate from their firms, who are allowed to roll over their retirement funds, to continue to enroll in plans with advantageous rates of return.

REFERENCES

This page intentionally left blank.