American Enterprise Institute Report: 'Changes to Household Retirement Savings Since 1989'

WASHINGTON, Oct. 21 -- The American Enterprise Institute issued the following report entitled "Changes to Household Retirement Savings Since 1989" by resident scholar Andrew G. Biggs.

Here are the excerpts:

Abstract:

This report uses two new data sources to provide insights on the evolution of retirement savings over the past three decades and how future retirees may fare. First, the Federal Reserve's Survey of Consumer Finances and Distributional Financial Accounts (DFA) provide estimates of both household savings in retirement accounts and any benefits households accrued under a traditionally defined benefit pension. The DFA data show that, from 1989 through 2016, household retirement savings increased for every age, income, race, or educational group. This is true whether retirement savings are measured in inflation-adjusted dollars or as the ratio of retirement savings to their annual earnings. Second, I present previously unpublished projections from the Social Security Administration's Model of Income in the Near Term (MINT). The MINT model projects that future retirees will have retirement incomes as a percentage of pre-retirement earnings similar to those of current retirees.

* * *

Americans are concerned about retirement. Eighty percent of Democrats and 75 percent of Republicans and independents agreed in a 2019 survey that "America is facing a retirement crisis" (Oakley and Kenneally 2019). Only 57 percent of Americans in a 2019 Gallup (2018) survey expect to have a comfortable retirement--the highest recorded in the past 15 years. Just 25 percent of Gallup's respondents believed they were saving enough for retirement.

Americans' worries are partly understandable. It is difficult to know how much income people need in retirement or how much they should save today to get there. And if people reach retirement age with insufficient savings, they may not have many options available to address the shortfall. But Americans' retirement worries today are exacerbated by a cottage industry of researchers, interest groups, and media reports that reinforces the view that households face a "retirement crisis" of inadequate savings. It is no wonder Americans are worried.

Many reports attribute the purported decline in Americans' retirement security to the generations-long shift from traditional defined benefit (DB) pensions to defined contribution (DC) retirement accounts, including 401(k)s and individual retirement accounts (IRAs). For instance, a study published by the progressive-leaning Economic Policy Institute states, "The shift from pensions to account-type savings plans has been a disaster for lower-income, black, Hispanic, non-college-educated, and single workers, who together add up to a majority of the American population" (Morrissey 2019). Monique Morrissey concludes that the US "retirement system does not work for most workers," which in her view "underscores the importance of preserving and expanding Social Security, defending defined benefit pensions for workers who have them, and seeking new solutions for those who do not."

But this is not the first time we have heard such dire warnings. In a 2002 Economic Policy Institute book, economist Edward N. Wolff (2002) analyzed the retirement prospects of Americans born between 1934 and 1951. These households were age 47 to 64 in 1998, the year of Wolff's analysis. Like Morrissey today, Wolff concluded that the shift from traditional pensions to 401(k)-type retirement accounts had been highly detrimental to the retirement income security of the typical US household. Wolff projected that 18.5 percent of the 1934-51 cohort households would retire into poverty, an increase from prior birth cohorts. Wolff also projected that 56 percent of the 1934-51 birth cohorts would have retirement incomes below 75 percent of their incomes immediately preceding retirement, a commonly used benchmark for retirement income adequacy.

To avert these problems Wolff recommended strengthening coverage by DB pensions and finding other ways to increase retirement wealth. Yet in hindsight, we now know these predictions were not borne out: According to Bureau of Labor Statistics data, the poverty rate for the 1934-51 birth cohorts as of 2018 was not Wolff's projected 18.5 percent but just 9.4 percent, which was not an increase from prior cohorts of retirees but a decrease.1 Likewise, Peter J. Brady et al. (2017) used IRS data to analyze the retirement income replacement rates of Americans born from 1938 to 1944. Three years following Social Security claiming, only 21 percent of these cohorts had retirement incomes less than 75 percent of their incomes just before retirement, less than half the share that Wolff had feared.

In this report, I bring new data to bear on the question of Americans' level of preparation for retirement. I first examine how Americans' retirement savings have evolved over the past three decades, analyzing household savings levels by age, income, race, and educational attainment. I then turn to a sophisticated retirement income projection model developed by the Social Security Administration (SSA) to assess the model's estimates of retirement income adequacy for present and future retirees. This report uses two new data sources to provide insights on the evolution of retirement savings/1 over the past three decades and how future retirees may fare. The first data source is the Federal Reserve's Distributional Financial Accounts (DFA). While the DFA are based on the Survey of Consumer Finances (SCF), the DFA include the value of households' accrued benefits in traditional pension plans in addition to the SCF's measures of balances in retirement accounts. The DFA allow household retirement savings to be viewed based on age, income, educational attainment, and race. The DFA data show that, from 1989 through 2016, household retirement savings increased for every age, income, race, or educational group. This is true whether retirement savings are measured in inflation-adjusted dollars or as the ratio of retirement savings to their annual earnings. I touch briefly on how the financial market downturns of early 2020 associated with the COVID-19 virus may have affected the savings of households near or early in retirement.

The second source of information is previously unpublished projections from the SSA's Model of Income in the Near Term (MINT), a microsimulation model of the US population that accounts for a wide range of income sources in retirement. The MINT model projects that future retirees will have replacement rates (retirement incomes as a percentage of preretirement earnings) similar to those of current retirees, whether measured in median replacement rates or the share of retirees with placement rates below 75 percent.

Together, the best data on retirement savings and the more sophisticated projections of future retirement incomes present a much more positive picture of US households' preparation for retirement than is commonly found in media reports or public opinion polls. Retirement policy should focus on filling existing gaps in retirement saving adequacy rather than large changes such as expanding Social Security benefits across the board.

What Has Happened to Retirement Savings Since 1989?

This section relies on two closely related sources of data on household retirement savings, the Federal Reserve Board's DFA and SCF. The DFA are built on the SCF, which is a household survey that focuses on Americans' assets, debts, savings, and other financial issues.

The SCF is commonly used to analyze Americans' retirement savings. It contains information on the balances of households' retirement savings accounts, such as 401(k)s and IRAs. But what differentiates the DFA from the SCF and from other household surveys is that the DFA contain estimates of households' accrued benefits under traditional DB pension plans./2

Most household surveys, including the SCF, collect data only on whether a household member participated in a traditional pension but provide no detail on the benefits the participant has accrued to date. That distinction is important, because simply participating in a traditional pension does not mean an individual accrued a significant future retirement benefit. Some workers will accrue nothing at all, because they changed jobs before being vested in the plan./3

Other pension participants will accrue relatively little because they failed to work a full career under the plan. Traditional pensions are back-loaded, meaning that for many years after beginning employment, participants accrue relatively few benefits. /4

Still other pension participants will accrue significant benefits by participating in a traditional pension throughout their working career. Due to these complicating factors, simple "participation" in a DB pension system is not comparable to participation in a 401(k)-style retirement account in which balances are easily measured and portable between jobs.

While traditional pensions are far less common in the private sector today, nearly all public-sector employees continue to participate in such plans. And public-sector employees in federal, state, and local government make up about 15 percent of the US workforce. Moreover, the DFA inclusion of accrued traditional pension benefits allow for a better view of the pension landscape as retirement plans shifted from traditional DB pensions to 401(k)-style retirement accounts. Thus, figures on accrued DB pension benefits are crucial for an accurate accounting of US households' retirement savings.

The DFA construct a single measure, called "pension entitlements," which merges various retirement-related financial assets. Pension entitlements include:

* The present value of benefits accrued under DB pensions;

* The balances of employer-sponsored retirement accounts, such as 401(k)s and 403(b)s; and

* The value of annuities and other retirement-related insurance contracts.

However, pension entitlements do not include balances in IRAs, much of which were rolled over from 401(k) plans or from lump-sum payouts from DB pensions.

To capture retirement savings in IRAs, I combine the DFA measure of pension entitlements with the SCF measures of IRA and Keogh account balances. While some higher-income households save for retirement outside of tax-favored plans, most households typically hold the vast majority of their retirement savings as either accrued traditional pension benefits, 401(k) or other employer-sponsored retirement accounts, or IRAs. Thus, combining the DFA measure of pension entitlements with IRA balances should provide a reasonable measure of the level and trend of household savings available to fund living expenses in retirement.

The DFA data allow retirement savings to be measured along several different group categories. These groups include:

* Age. Age groups include households under age 40, age 40-54, age 55-69, and age 70 and over.

* Income. Income groups are broken down by fifths, or quintiles, of the income distribution.

* Race. Racial groups include white, non-mixed race; Hispanic or Latino, non-mixed race; Black or African American, non-mixed race; and other.

* Education. Educational groups include less than high school education, high school diploma or GED, some college, and bachelor's degree or more.

Unfortunately, the DFA data do not allow for cross tabulations across groups, such as income by age. A currently underway project that will calculate pension entitlements directly from SCF data should allow for that greater level of detail.

The DFA provide quarterly data on retirement savings for each subgroup from 1989 through 2016. Data on the number of households in each group and the annual earnings of such households are drawn from the SCF, which is conducted every three years.

From these data, I produce two figures to summarize household retirement savings. The first is average household retirement savings in dollar terms, adjusted for inflation to 2016./5

This measure shows the real resources available to households and helps gauge the absolute standard of living they may enjoy in retirement. The second is household retirement savings as a percentage of annual household earnings. This measure comparing savings to salaries helps gauge how well households' savings can replace their earnings once the household retires from the workforce. /6

Together, these measures provide a broader view of household retirement savings than is available in most other datasets./7

However, before turning to results, I first discuss the important role of Social Security in furnishing income in retirement.

* * *

About the Author

Andrew G. Biggs is a resident scholar at the American Enterprise Institute. He previously served as the deputy commissioner of the Social Security Administration. He currently serves as a member of the Financial Management and Oversight Board for Puerto Rico.

* * *

REPORT and FOOTNOTES: https://www.aei.org/research-products/report/changes-to-household-retirement-savings-since-1989-2/