The U.S. Government Accountability Office spent a year reviewing alternative retirement plan designs abroad and proposals for new plan designs in the United States to determine which features in these plans could potentially address retirement risks currently faced by American workers. GAO modeled the impact on retirement income potential from defined contribution plans in three scenarios: mandated universal access, automatic enrollment, and mandated employee participation.
GAO found that universal access and mandated employee participation together would increase the share of workers with some defined contribution plan savings at retirement from a current 67 percent level to 97 percent. It would raise the equivalent annuity benefit from a baseline of $15,217 a year to $21,312 a year. In a scenario where there is mandated access and automatic enrollment, but no required employee participation, the results are almost as impressive: the share of workers with savings at retirement rises to 91 percent with an $18,556 annual annuity equivalent.
By Robert Stowe England
August 26, 2009
The U.S. Government Accountability Office has completed an ambitious year-long research project that sought to identify the retirement security risks that American workers increasingly face and looked to find retirement plan designs and features that could help address those risks.
The GAO released the results of its efforts in a 72-page report August 24. The report was completed in late July, but its release was withheld for 30 days by Representative George Miller (California Democrat), Chairman of the House Committee on Education and Labor, who requested the study.
The report is titled Private Pensions: Alternative Approaches Could Address Retirement Risks Faced by Workers but Pose Trade-offs and can be found at this link:
The GAO report finds a number of worrisome trends. For one thing, the share of Americans working in the private sector earning a defined benefit pension – one that pays an annuity income for life at retirement based on their earnings – has declined as employers have terminated or frozen those plans.
Workers increasingly rely on defined contribution plans, which usually pay a lump sum at retirement and rely mostly on employees to save for their own retirement and to make investment choices from among options in a plan.
Another worry is that many workers who are saving through 401(k)’s and other vehicles are not accumulating sufficient funds to be cover their needs and living expenses in retirement.
To bolster its case, GAO cites a 2007 study by Alicia Munnell and others that the median combined balances in 401(k) plans and Individual Retirement Accounts (IRA) totaled $78,000 for people aged 55 to 64. (See Footnote 1.) The 2008-2009 market crashed decimated many of those balances.
Further, older workers in retirement savings plans have failed to move assets away from riskier equities to bonds as they age. A study by Jack VanDerhei, for example, found that nearly one in four people between the ages of 56 and 64 had more than 90 percent of their account balance invested in stocks at the end of 2007. (See Footnote 2.)
GAO identified a range of problems with current retirement savings patterns and arrangements.
• Too few workers are covered (only 50 percent of the private sector workforce).
• Contributions are inadequate.
• Workers face investment risk.
• Benefits are not always portable and when workers change jobs, they often take pre-retirement withdrawals.
• Workers borrow from saved funds or take hardship withdrawals.
• Administrative and investment fees can sometimes eat away much of the investment return.
• Workers tend to draw down benefits early in retirement, leaving too little for later years.
• Retirees continue to face investment risk from the assets in a retirement savings plan.
• Retirees also face inflation risk, as the purchasing power of savings or annuities declines over time.
Barbara Bovbjerg, director of education, workforce and income security at GAO, was in charge of the study and a research team working for her examined the design, features and ultimate benefits provided by private pension schemes abroad. They also reviewed four domestic proposals for new retirement plan designs in the United States.
The GAO researchers sought to identify certain design features of retirement plans that, if adopted, could yield a more adequate benefit in retirement.
GAO reviewed the private sector schemes in the United Kingdom, the Netherlands and Switzerland. They found a number of features in these plans that boosted retirement income and security.
In the United States, the GAO examined the Urban Institute’s Super Simple Saving Plan, the ERISA Industry Committee’s New Benefit Platform for Life Security, The New America Foundation’s Universal 401(k) Plan, and The Economic Policy Institute’s Guaranteed Retirement Accounts Plan.
The domestic proposals were selected, the GAO report states, “because they incorporate strategies to address risks workers face, are developed in enough detail to allow us to fully analyze them, are not duplicative, and have been proposed or considered in the last five years.”
“Neither the pension systems in other countries we reviewed, nor the domestic proposals constitute a panacea for the challenges of the U.S. pension system,” GAO concludes.
“No system or proposal is perfect and each requires careful consideration of the trade-offs between its advantages, costs, and responsibilities,” the GAO report states.
Nevertheless, GAO found that the foreign pension schemes have either yielded impressive results for workers, as in the case of the Netherlands and Switzerland, or promise to significantly improve retirement income, as in the case of the United Kingdom.
“Despite important social, economic, and institutional differences between the United States and these countries, key features from these models, as well as the domestic proposals, are relevant and could potentially offer some solutions for the U.S. pension system,” the report stated.
“Taken together, these key features could be used more comprehensively [to] address risks workers face,” GAO stated. “The challenge for Congress will be to balance the interests and responsibilities of workers, employers, and the government and find the most promising steps to help Americans achieve retirement security.”
Three Modeled Scenarios
For its simulation, the GAO paper has focused on those options that are most likely to increase coverage rates and the amount of the final benefit.
GAO ran its three scenarios through the Pension Simulator (PENSIM), a software model developed to analayze lifetime coverage and adequacy of employer sponsored plans. It was developed by the Policy Simulation Group in Washington, D.C., through the sponsorship of the Employee Benefits Security Administration at the Department of Labor and is one of three simulation models developed by PSG.
The simulations were based on a sample of the 1990 cohort, totaling 126, 518 at birth. GAO, which has the PENSIM software in-house, simulated the timing of lifetime events for this cohort, as well as their future savings through employer-sponsored defined contributions plans.
The three options for which modeled by the Pension Simulator are as follows:
• Universal access. All employers that do not sponsor a plan are required to provide a defined contribution plan with no employer contribution. Existing employer-sponsored plans are not affected.
• Universal access with automatic enrollment. In addition to universal access, as described above, all defined contributions plans have automatic enrollment in which individuals must affirmatively opt out of participating in the plan.
• Universal access with mandatory participation. In addition to universal access, all workers with access to a defined contribution plan are required to participate.
The contribution levels -- an important factor in the outcome of the three scenarios -- is not not described in the report. A description of how PENSIM calculates contribution rates in defined contribution plans has been provided to Mind Over Marekt by Martin Holmer at the Policy Simulation Group and is provided below in Footnote 3.
Basically, the simulations are based on the recent historical patterns of contribution rates in defined contributions plans and includes both employee contributions and employer matches. PENSIM randomly generates contribution amounts and matches for the sample from the 1990 birth cohort.
Employees are assumed to have been invested in life cycle funds, which reduce the level of equity exposure over time as workers near retirement.
The simulations assume a real rate of return of 2.9 percent on investments in participants' accounts.
GAO found that requiring universal access where participation is voluntary increases the share of workers with defined contribution savings at retirement from 67 percent to 79 percent.
The increase in coverage is greatest for low-income workers, where the share with some defined contribution savings rises from 48 percent to 63 percent.
Requiring universal also boosts retirement income, GAO found. Using a measure of the annuity equivalent of the accumulated balance in the plan, GAO found that average household benefit would rise 12 percent – from a baseline of $15, 217 to $17,058 a year in constant 2008 dollars.
Adding automatic enrollment to universal access yielded even better results. Under this scenario 91 percent of workers have defined contribution savings at retirement. For low-income workers in the bottom quartile of earnings, the coverage rate would rise from 653 percent under universal access alone to 84 percent.
Overall, the projected average pension income would increase to $18,556 from the baseline $15,217.
Requiring all private sector workers to participate in a defined contribution plan yields the “largest overall gains compared to the baseline,” GAO concludes.
The share of workers with defined contribution savings at retirement rises to 97 percent. For low-income workers, it rises to 96 percent, which is 14 percentage points higher than simply providing universal access.
The average income would rise to $21,312, compared to the $15,217 baseline. For the low income quartile, income would rise to $5,157 from $2,761 in the baseline scenario.
The model scenarios yield coverage rates and benefits that are more in line with some of the benefits to be found in the foreign pension schemes reviewed in the GAO study.
In the Netherlands, the predominant plan is a career-average defined benefit plan and coverage and contributions are mandatory for most employers and workers. Employers contribute 7 to 19 percent of pay, while workers contribute 3 to 8 percent of pay. Not surprisingly 90 percent of workers are covered. The goal for this plan is to replace 70 percent of pre-retirement pay, when combined with a basic Social Security plan.
In Switzerland, the predominant retirement benefit is a cash balance plan (a hybrid of the defined benefit plan), and coverage is mandatory with 90 percent of workers covered. Both employers and employees are each required to contribute 3.5 to 9 percent of pay. The contribution levels rise with age. The goal for this plan is to replace 60 percent of pre-retirement pay, when combined with Social Security.
In the United Kingdom, there is a new Personal Accounts plan, which consists of defined contribution plans. When fully phased in, employers will contribute 3 percent of pay, employees will contribute 4 percent of pay, and the government will contribute 1 percent of pay. The goal for this approach is to replace 45 percent of pre-retirement income, when combined with Social Security.
In all three countries the majority of the benefit, usually 75 percent, must be taken as an annuity to ensure that retirees are less likely to outlive their retirement funds.
Copyright © 2009 Robert Stowe England
Footnote 1. Alicia Munnell, Francesca Golub-Sass, and Dan Muldoon, “An Update on 401(k) Plans: Insights from the 2007 Survey of Consumer Finances,” Center for Retirement Research at Boston College, March 2009.
Footnote 2. Jack VanDerhei, “The Impact of the Recent Financial Crisis on 401(k) Account Balances,” Employee Benefit Research Institute, Issue Brief, no. 326 (Washington, D.C., February 2009)
Footnote 3. It would seem, based on the Note to Table 12 on page 53 and to Table 13 on page 63) that GAO used the baseline PENSIM assumptions about defined contribution plan contribution rates, according to Martin Holmer at the Pension Simulation Group in Washington, D.C. The baseline contribution assumptions in PENSIM are based on employee contribution behavior, observed in the EBRI-ICI database of 401(k) participants at the end of the 1990s, and on employer matching behavior, observed in the late 1990s BLS Employee Benefit Survey, which is now called the National Compensation Survey. The details about how this information is incorporated into PENSIM is thoroughly documented in the PENSIM Overview document, which is available on the Documentation page of the Policy Simulation Group web site at this link: http://www.polsim.com/overview.pdf. The PENSIM Overview discusses employee contribution rates in three sections (going from general to specific), which are on pages 14-15, pages 145-146, and pages 224-225. In the EBRI-ICI data, the employee contribution rate varies widely by age and earnings level and for unknown reasons. See, for example, the chart on page 5 of the EBRI Issue Brief 238. In addition to this variation among those who make a positive contribution, a small group of plan participants make zero contributions in a year. This behavior is simulated in PENSIM along with the wide variation among those with positive contributions. PENSIM simulates in detail all the federal rules that limit high contributions. And remember that while the dollar limits are inflation indexed, after many decades the limits will be lower when measured relative to earnings (which are projected to grow faster than prices). This is relevant because GAO is simulating a sample of individuals born in 1990, whose peak earnings years will be during the 2040s and 2050s. Martin Holmer reports that the Policy Simulation roup has conducted a number of validation tests of defined contribution plan balances simulated by PENSIM. In comparisons with both data from EBRI-ICI and from the 2004 Survey of Consumer Finances, we find PENSIM simulated defined contributionC balances are quite similar to those observed in real-world data. For more on those validation tests, see PENSIM Overview, Chapter 10.