![U.S. flag](https://www.dol.gov/themes/opa_theme/img/flag-favicon-57.png)
An official website of the United States government
The 20th century was the dawn and expansion of retirement security for American workers, with the introduction of private pension plans, the enactment of Social Security, court rulings allowing collective bargaining of employee benefits, and the enactment of several federal laws governing pension plans. The passage of the Employee Retirement Income Security Act (ERISA) in 1974 was among one of the most impactful events; ERISA established guardrails for employer-provided retirement benefits. Fifty years later, the retirement landscape looks very different, with new types of plans and a shift toward greater worker responsibility for their own retirement security. Paralleling this shift over the last half century, the Bureau of Labor Statistics (BLS) tracked changes in retirement benefits and costs. In this article, I look at how the retirement landscape has changed since ERISA, how BLS has tracked those changes, and where the system stands today.
Defined benefit and defined contribution are the two main categories of retirement plans, although in each case there are variations in plan structure and terminology. In general, a defined benefit plan guarantees an employer-provided retirement benefit that an employee will receive for life, typically based on earnings and years of service. In a defined contribution plan, employees have an individual retirement account that may be funded by both employee and employer contributions. Also, in a defined contribution plan, benefits are not guaranteed and can fluctuate because of investment returns. As I will discuss, the prevalence and features of these plans have changed over the past 50 years, as has the information provided by BLS.
BLS gathered information on employee benefits periodically throughout the first half of the 20th century, including special studies and questions added to wage surveys.1 Starting in the late 1950s, BLS added studies of employer costs for benefits (more on costs below) and a digest that detailed the provisions of retirement plans from specific companies. A 1974 study of a representative sample of defined benefit plans covering 100 workers or more provided a benchmark of the landscape just prior to ERISA.2 While many of the plans studied included provisions that met the ERISA standards, some plans had more stringent requirements. Here are a few examples:
While ERISA does not require private employers to provide retirement benefits, the law imposes requirements when such plans are provided. Although ERISA provisions affect both defined benefit and defined contribution plans, much of the initial focus was on defined benefit plans, which were the predominant type of retirement plan in 1974. Included among the provisions of ERISA are reporting and disclosure requirements (providing information on plan operations to participants and the Department of Labor), funding and fiduciary standards (ensuring that defined benefit plans have sufficient assets that are invested in the best interest of participants), and an insurance program to protect defined benefit plan participants (through the Pension Benefit Guaranty Corporation).4 ERISA also required that defined benefit plans meet certain standards to obtain favorable tax status. Here are some examples of ERISA provisions:
BLS collects information about these retirement provisions, which are discussed in detail below.
While not related to the passage of ERISA, in the late 1970s, BLS introduced a program to report on the incidence and detailed provisions of employee benefits; this included retirement plans. First, data on private sector benefits were collected in conjunction with information on wages by occupation for use by the Office of Personnel Management in making comparisons between federal and nonfederal compensation. Ultimately, only wage comparisons were used in federal pay calculations, which are currently governed by the Federal Employees Pay Comparability Act of 1990.5 While not used for compensation comparison purposes, BLS continues to capture and report on the details of employee benefits.
Initial BLS benefit surveys focused on full-time workers in large private sector establishments. Beginning in 1979, BLS published annual estimates of the incidence and the provisions of employee benefits, including retirement plans. Among medium and large private firms in 1979, 87 percent of full-time workers participated in a retirement plan––typically a defined benefit plan. As reported by BLS in “Employee benefits in industry: a pilot survey” and subsequent reports, a typical private sector defined benefit plan in the late 1970s and early 1980s, consistent with ERISA provisions, included the following features:
BLS continued to track retirement benefits throughout the 1980s, first focusing on large private employers and then adding small private employers and state and local governments. Small employers were less likely to provide defined benefit plans than their larger counterparts, while state and local governments frequently provided such benefits. Chart 1 shows the share of employees, in state and local government and private industry, participating in retirement plans in 1989 and 1990.7
The information on small private employers and state and local governments revealed differences in benefit amounts and other provisions for different groups of workers. For example, nearly all defined benefit plans for state and local government workers computed benefits as a percentage of earnings times years of service, while some private sector plans used dollars-times-years formulas. Typical state and local government plans provided 2 percent or more for each year of service. In contrast, private sector plans (regardless of employer size) provided about 1.5 percent for each year of service. In some cases, state and local government workers were not covered by Social Security, which may account for some of the differences in benefit provisions.8
In the 1990s, BLS expanded its benefit studies to provide greater detail. Information became available by industry, with greater occupational detail, and for several new groupings, including workers who are covered by a collective bargaining agreement, full-time workers or part-time workers, wage level, and broad geographic areas. The expanded dataset identified differences in the availability of defined benefit plans for different groups of workers, as well as the relationship between groups. For example, the higher share of workers participating in retirement plans in certain industries or locations may be related to a higher concentration of union workers. (See chart 2.)
ERISA did not ignore defined contribution plans, but such plans may have received less attention in the 1970s because these plans were not as prevalent as defined benefit plans. Nevertheless, ERISA did impose participation and vesting requirements, which were similar to the requirements for defined benefit plans. But it was another law, the Revenue Act of 1978, that helped promote the use of defined contribution plans in the 1980s and beyond with the addition of section 401(k) to the tax code. This new feature allowed workers to defer paying tax on earnings saved in an employer’s plan. BLS tracked the shift from defined benefit to defined contribution plans throughout the latter part of the 20th century, which is evident when comparing results among full-time employees in private establishments with 100 workers or more. (See chart 3.)
While defined contribution plans can take many forms, the most prevalent plans were categorized by BLS as savings and thrift plans, which generally required employee contributions that were matched by the employer. As initially captured by BLS in the late 1980s through the 1990s, savings and thrift plans had the following typical features:
As employers continued to alter their retirement plans, BLS continued to identify what new information to capture and how best to report on current plan arrangements. For example, the rise of defined contribution plans meant that employees often had to make a positive decision to join the plan, as opposed to automatic participation in a defined benefit plan (once meeting age or service requirements). To gain a clearer picture of whether employees chose to join the plan, BLS provided new information that identified the share of workers who were offered defined contribution plans and those who participated in a plan. Such information helped to illuminate variations in the “take-up rate,” which is the share of employees who chose to participate when offered a plan. These data also helped to track whether changes in plan features, such as automatic enrollment, influenced take-up rates. Chart 4 illustrates that an increasing share of workers were offered a defined contribution plan throughout the 2000s. Participation rates remained relatively stable for many years but have shown recent increases.
In addition, changes in retirement plans made it necessary for BLS to gain a better understanding of whether defined contribution plans were offered as a replacement for defined benefit plans or whether employees now had both plans available. When first introduced in the 1980s, many employers added 401(k) plans as an addition to their existing defined benefit plan. But, over time, the share of workers participating in defined benefit plans continued to decline while overall retirement plan coverage remained stable. Intuitively, this movement meant that fewer workers had both plans. But how best to tabulate and express this change?
Throughout the 1990s, BLS provided separate data for those who participated in any retirement plan, in a defined benefit plan, and in a defined contribution plan. Through a little math manipulation, some sense of the share of workers with both plans is possible but with a large degree of uncertainty. Throughout the 1990s, the estimated share of workers with both plans was consistently around 30 percent. The share with only a defined benefit plan declined from 30 percent to 22 percent over the decade, and there was a nearly identical rise in the share of employees with only a defined contribution plan. (See chart 5.)
To gather more precise information on retirement plan participation, BLS began asking employers to provide counts of workers with various combinations of retirement plans in the late 1990s. This proved challenging because employers often did not keep records in the desired manner and were not always able to provide separate counts of workers in one plan versus workers in a combination of plans. Despite these data collection challenges, BLS was able to publish greater detail on retirement plan combinations, which confirmed the trend toward sole coverage by a defined contribution plan. By 2023, only 3 percent of all private industry workers had access to only a defined benefit plan, 12 percent had access to both plans, and 56 percent only had access to a defined contribution plan.11
Another challenge related to counting the share of workers with certain benefits is the trend toward employers “freezing” their defined benefit plans. Employers can terminate their plan and use assets to provide benefits to plan participants, often through the purchase of an annuity. Following such a termination, BLS would report zero workers with a defined benefit plan at a company. But other variations on plan freezes can be more problematic to track. In some cases, employers continue the defined benefit plan for existing participants but do not allow any new participants. Here, BLS includes only those workers accruing benefits as plan participants. Alternatively, employers may maintain the defined benefit plan for existing participants but without accruing any further benefits. Currently, such workers are counted as offered a plan and participating in a plan, and additional information is published that shows the share of employees who no longer accrue benefits. In addition to tracking the declining availability of defined benefit plans in the private sector, recent BLS data show that about 40 percent of participants in a defined benefit plan were in plans that had instituted some type of freeze.12
Generally, capturing, tabulating, and reporting on retirement plan features, whether defined benefit or defined contribution, continue to pose challenges because of the wide variety of provisions. Unlike other BLS statistics, such as the unemployment rate, the Consumer Price Index, or productivity measures, the details of retirement benefits (and, in fact, most employee benefits) do not lend themselves to a single number. And the more detail that is displayed, the more difficult it may become to understand the key features of a plan.
Consider the various options that have been used to identify the age and service requirements for normal retirement in defined benefit plans. Table 1 shows the traditional BLS approach to presenting this information, which is to provide every possible variation. The original publication of these tables even includes footnotes that identified rare features not specifically included in the table.13 Providing this amount of detail allows data users to find the most prevalent provisions, track whether their plan is similar to others, and identify any changes in plan features over time. If the main takeaway from table 1 is that normal retirement is frequently available at age 65, but many plans allow normal retirement at earlier ages with sufficient service, then all the additional detail may not be needed.
Characteristic | Percent |
---|---|
Total | 100 |
No age requirement | 11 |
30 years of service | 11 |
Age 55 or less | 8 |
20 years of service | 6 |
30 years of service or more | 3 |
Age 56 to 60 | 10 |
No service requirement | 2 |
5 years of service | [1] |
10 to 15 years of service | 4 |
20 to 25 years of service | 2 |
30 or more years of service | 3 |
Age 61 to 64 | 19 |
No service requirement | 4 |
5 years of service | 1 |
10 to 15 years of service | 11 |
20 to 25 years of service | 2 |
30 or more years of service | 1 |
Age 65 | 45 |
No service requirement | 39 |
5 years of service | 2 |
10 to 15 years of service | 4 |
Sum of age plus service | 6 |
Equals less than 80 | 1 |
Equals 80 | 1 |
Equals 85 | 3 |
Equals 90 or more | 1 |
[1] Indicates less than 0.5 percent or data not available. Note: Data are for full-time employees participating in all defined benefit plans, medium and large private establishments, 1980. Source: U.S. Bureau of Labor Statistics. |
Table 2 shows a newer approach, which attempts to simplify the presentation and focus on the key points. In this table, the array of ages shown is reduced to two—age less than 65 and age 65—and the service requirements are similarly reduced. These entries are complemented by information on the median values—age 65 with 5 years of service. The takeaway in this case is largely the same as in the first table. Is this an improvement? Is the effort to capture, review, and tabulate the variety of details from thousands of plans worth the result? These are important questions to answer since part of the mission of BLS, as a fact-finding agency, is to provide reliable and clear information to the public. This may be a case where artificial intelligence could simplify the process and provide a clearer answer. In fact, BLS currently uses a variety of machine learning approaches to review and synthesize text and has experimented with such approaches on descriptions of employee benefits. As these processes mature, these approaches hold the potential to quickly distill the highlights of retirement plans.
Characteristic | Percent |
---|---|
Total | 100 |
With age and service requirement | 80 |
With age only requirement | 17 |
With service only requirement | [1] |
With age plus service requirement | [1] |
Total | 100 |
Age less than 65 | |
Less than 10 years of service | 13 |
10 years or more of service | 28 |
Age 65 | |
Less than 10 years of service | 55 |
10 years or more of service | [1] |
Age requirement | |
10th percentile | 60 |
25th percentile | 62 |
50th percentile (median) | 65 |
75th percentile | 65 |
90th percentile | 65 |
Service requirement | |
10th percentile | 5 |
25th percentile | 5 |
50th percentile (median) | 5 |
75th percentile | [1] |
90th percentile | 25 |
[1] Indicates no data were reported or data do not meet publication criteria. Note: Data are for all employees participating in traditional defined benefit plans, all private establishments, 2019. Source: U.S. Bureau of Labor Statistics. |
In addition to looking at plan incidence and provisions, BLS tracks the employer costs of providing benefits to employees. In the Employer Expenditures for Employee Compensation survey that BLS produced from 1959 to 1977, BLS provided average employer expenditures for individual benefits. In 1966, private employers spent 2.6 percent of their compensation expenses on retirement benefits, nearly all of which was for defined benefit plans. And while the share of private employer compensation costs spent on retirement plans increased to 4.3 percent by 1977, defined contribution plans remained a minor part of the cost.14
In 1975, BLS introduced the Employment Cost Index (ECI), which––in parallel to the Consumer Price Index––tracks the cost to employers of a “basket” of workers. BLS identifies a representative sample of occupations across industries and captures employer costs for their wages and benefits over time. While the ECI is an important gauge of wage and benefit inflation, the source data are also used to show actual employer costs. These costs are represented as dollars per hour worked and the share of employer costs spent on individual benefits. This series, known as Employer Costs for Employee Compensation, began providing data on individual benefits in the mid-1980s. In 1987, private employers spent 48-cents-per-hour worked on retirement plans, with most (42 cents) going toward defined benefit plans (then referred to as simply “pension” plans) and the remainder (6 cents) going toward defined contribution plans (then referred to as “savings” plans). For the past 35 years, private employers have consistently spent between 3 percent and 4 percent of their compensation expenditures on retirement benefits, but, reflecting the shift in plan type, much of that cost is now spent on defined contribution plans. (See chart 6). In December 2023, employers spent about 70 percent of their retirement expenditures on defined contribution plans.15
Change is the only constant when it comes to retirement plans in the past half century. Several features that traditionally were unique to defined benefit plans are now becoming common in defined contribution plans and vice versa. Here are a few examples:
In the half century since the passage of ERISA, the retirement landscape has changed dramatically. ERISA has been amended multiple times, generally to make it easier for employees to obtain retirement benefits. At the same time, employers continue to introduce alternative retirement arrangements that can both attract and retain workers while controlling costs. Today, defined benefit plans are rare in the private sector, with new hybrid variations that resemble defined contribution plans. The now predominant defined contribution plans incorporate newer features such as automatic enrollment and escalation, Roth contributions, investment in lifecycle funds, and the option to turn account balances into a lifetime annuity. BLS will continue to track these changes and strive to meet the challenges of how best to capture and describe plan coverage and key plan features.
William J. Wiatrowski, "ERISA at 50: BLS tracks the evolution of retirement benefits," Monthly Labor Review, U.S. Bureau of Labor Statistics, July 2024, https://doi.org/10.21916/mlr.2024.13
1 For several decades in the mid-20th century, the Bureau of Labor Statistics (BLS) produced a series of wage studies, including Community Wage Surveys, Area Wage Surveys, and Industry Wage Surveys. These surveys often included a few questions about the availability of employee benefits, including retirement plans. BLS also produced collections of benefit plan features, such as various digests of selected pension plans. Many of these publications can be found in the FRASER library of economic history, which is maintained by the Federal Reserve Bank of St. Louis, see "Bureau of Labor Statistics," FRASER, Federal Reserve Bank of St. Louis, https://fraser.stlouisfed.org/author/united-states-bureau-labor-statistics.
2 John W. Thompson characterized this 1974 study as follows: “The data used in this article are from a 1974 BLS analysis of a probability sample of employment-related basic defined benefit plans in private industry, whose administrators reported to the Department of Labor in accordance with the Welfare and Pension Plan Disclosure Act of 1959, as amended.” John W. Thompson, “Defined benefit plans at the dawn of ERISA,” Compensation and Working Conditions (U.S. Bureau of Labor Statistics, March 30, 2005), p. 6, https://www.bls.gov/opub/mlr/cwc/defined-benefit-plans-at-the-dawn-of-erisa.pdf.
3 Ibid.
4 The role of BLS is to gather and disseminate information on labor-related statistics, including employer-provided retirement plans. BLS is not the legal expert on the Employee Retirement Income Security Act (ERISA) or related laws. Detailed information on ERISA is available from a variety of sources. The Department of Labor’s Employee Benefits Security Administration provides information on ERISA, see “History of EBSA and ERISA,” Employee Benefits Security Administration (U.S Department of Labor), https://www.dol.gov/agencies/ebsa/about-ebsa/about-us/history-of-ebsa-and-erisa. The Congressional Research Service also provides a detailed summary of ERISA, see Patrick Purcell and Jennifer Staman, “Summary of Employee Retirement Income Security Act (ERISA),” RL34443 (Congressional Research Service, May 19, 2009), https://crsreports.congress.gov/product/pdf/RL/RL34443/6. A timeline on changes in retirement legislation is available from Georgetown University Law Center, see “A timeline of the evolution of retirement in the United States” (Georgetown University Law Center, 2010), https://scholarship.law.georgetown.edu/cgi/viewcontent.cgi?article=1049&context=legal. Readers interested in the Pension Benefit Guaranty Corporation should see https://www.pbgc.gov/.
5 For information on H.R.3979––Federal Employees Pay Comparability Act of 1990, see "H.R.3979––Federal Employees Pay Comparability Act of 1990," 101st Congress (1989–1990), https://www.congress.ogv/bil/101st-congress/hous-bill/3979".
6 “Employee benefits in industry: a pilot survey,” Report 615 (U.S. Department of Labor, Bureau of Labor Statistics, July 1980), https://www.bls.gov/ebs/publications/pdf/report-615-july-1980-employee-benefits-in-industry-a-pilot-survey.pdf. The BLS 1979 survey of employee benefits included just a few defined benefit plan provisions. Additional detail was added in subsequent years. Examples shown are from the BLS benefit surveys conducted from 1979 to 1983.
7 Information for small private employers and state and local governments was collected in 1990, and information for medium and large private employers was collected in 1989. These studies and others are available on the BLS website, see “Annual summaries of benefit coverage,” Employee Benefits (U.S. Bureau of Labor Statistics), https://www.bls.gov/ebs/publications/annual-benefits-summary.htm.
8 When first enacted, Social Security generally did not cover public sector employees, who were often covered by a defined benefit plan. Because such plans were the only source of retirement income for public sector employees, they often provided more generous benefits than private sector plans.
9 While posttax contributions were allowed by many plans, these were typically a legacy of pre-401(k) savings plans. Such provisions declined over time but reemerged following the introduction of Roth 401(k) contributions, which became effective in 2006. In such plans, employee contributions are posttax, but distributions are tax free when withdrawn.
10 See “Annual summaries of benefit coverage,” https://www.bls.gov/ebs/publications/annual-benefits-summary.htm.
11 “Employee benefits in the United States, March 2023,” Employee Benefits (U.S. Bureau of Labor Statistics, last modified September 21, 2023), https://www.bls.gov/ebs/publications/employee-benefits-in-the-united-states-march-2023.htm.
12 Ibid.
13 For example, some plans had two or more age and service requirements, such as a lower age with more service and a higher age with less service. And some plans had unusual service requirements not shown separately, such as 7 or 8 years.
14 William J. Wiatrowski, “Tracking changes in benefit costs,” Compensation and Working Conditions (U.S. Bureau of Labor Statistics, Spring 1999), https://www.bls.gov/opub/mlr/cwc/tracking-changes-in-benefit-costs.pdf.
15 Employer Costs for Employee Compensation––December 2023, USDL-24-0485 (U.S. Department of Labor, March 13, 2024), https://www.bls.gov/news.release/archives/ecec_03132024.htm.
16 To learn more about cash balance plans, see “Fact sheet: cash balance pension plans,” Employee Benefits Security Administration (U.S. Department of Labor, November 2011), https://www.dol.gov/agencies/ebsa/about-ebsa/our-activities/resource-center/fact-sheets/cash-balance-pension-plans.
17 For both defined contribution and cash balance defined benefit plans, separating employees have options, with corresponding tax consequences, for how to handle available funds.