Older Workers Report

Recession Increases Downward Mobility in Retirement: Middle Earners Hit From Both Sides

May 17, 2020

Older Workers at a Glance...

  • 13.6% U-3 Headline Unemployment
  • 63.9% Older Workers’ Labor Force Participation
  • 6.8 M Older Workers In Frontline Jobs
  • 40% Older Workers Without Paid Sick Leave
  • Increased Downward Mobility At All Earnings Levels: An additional 3.1 million older workers will fall into lifelong poverty in retirement. Overall, the 67 million older workers and their spouses in the U.S. will suffer a decrease of 7 percentage points in their retirement replacement rate.click
  • Middle Earners Hit Twice: Middle earners sustain both job loss and market loss, leading to an additional 1.1 million older workers falling from the middle class into poverty.
  • Policy Recommendations: Older workers were left out of the government response to the COVID-19 recession. In the short term, Congress should discourage early retirement withdrawals and increase and extend unemployment benefits for older workers. The recession exposes the need for comprehensive reform: expanding Social Security and creating a public option retirement plan in the form of Guaranteed Retirement Accounts.

Download the full report here. 


All Earnings Levels: Increased Poverty

Low Earners: From Bad to Worse

Low-earning older workers—those who earn less than $47,600, the median earnings for workers ages 50 to 60—are least likely to be able to work from home and therefore more likely to be unemployed in the pandemic’s aftermath. Twenty-nine percent of low-earners are expected to lose their jobs in the COVID-19 recession, compared to 25% for all older workers, (see Figure 4). When low-earning older workers are laid off, they withdraw what little they have saved for retirement (63% have nothing saved and the average among those with any savings is $42,000, see Figure A5) or they go into debt to make ends meet until they can claim Social Security, mostly at age 62.4

While a large majority of low-earning older workers were likely to be poor in retirement before the recession, the COVID-19 fallout will push an additional 1 million people from this group into poverty. The increase in poverty resulting from the COVID-19 recession is relatively low, with an increase from 87% to 90%. This is because, as mentioned earlier, most low earners have nothing saved for retirement and are likely to rely on Social Security (see figure A5). The COVID-19 recession is projected to reduce the replacement rate of all low-earning workers by 7 percentage points, falling to 56% from 63% (see Figure 2). To put this decrease into perspective, a household earning $40,000 a year would lose $2,800 in annual retirement income, falling to $22,400 from $25,200 before the recession.



Middle Earners: Hit From Both Sides

Our broken retirement system hits middle earners (households earning above $48,000 but below the Social Security earnings cap of $137,700) hardest because they are susceptible to both job loss and market loss. Twenty-two percent of older, middle-earning workers will likely lose their jobs in this recession, compared to 25% of older workers overall and only 15% of high-earning older workers (See Figure 4). Unlike low earners, middle earners have some retirement savings, with 41% reporting no retirement savings but an average of $101,000 among those who do have savings (see Figure A5). However, these funds will likely fall prey to financial market losses, and what remains is vulnerable to emergency withdrawals as households make up for lost earnings.5 This one-two punch is expected to leave more than half of middle earners who lose their jobs with less than $5,0006 in savings when they retire at 65.7

As a result, poverty rates for middle earners will increase to 42% from 38%, representing an additional 1.1 million people experiencing downward mobility—falling from the middle class into poverty—in retirement due to the recession (see Figure 1).

The recession will decrease all middle earners’ preparedness for retirement. Before the recession, these workers could expect to replace 62% of earnings with retirement income if they retire at 65. Post-recession, their replacement rate will fall 10 percentage points to 52% (see Figure 2). To put this decrease into perspective, a household earning $100,000 a year would lose $10,000 in annual retirement income, $52,000 compared to $62,000 before the recession. 


High Earners: Surprising Risk of Old-Age Poverty

While only 15% of high earners are expected to lose their jobs as a result of the COVID-19 recession (see Figure 2), their retirement is likely to take a bigger hit from market losses. Because Social Security provides a lower replacement rate for high earners (see Figure A2), they are more reliant on retirement savings to maintain their lifestyle and therefore more exposed to market risk. Even before the COVID-19 recession, many high-earning households were unlikely to maintain their living standards in retirement. Among high earners—those with earnings above the Social Security cap of $137,700—27% do not have any retirement savings. For those who do have retirement assets, the average savings projected by age 65 was $252,000, which would cover at most two years of retirement at the same living standard (see Figure A5). Post-recession, the average retirement account balance at 65 is projected to be $173,000, which is $79,000 less than projected without the recession.

High-earning older households are a small group, representing only 4 million out of 67 million people in older households. Yet, the COVID-19 recession will double the number of high earners expected to experience downward mobility into poverty in retirement, increasing from 360,000 to 720,000 people (see Figure 1).

While projected elder poverty rates are relatively low for higher earners both before and after the recession (4% before and 8% after), the drop in living standards is substantial. Downwardly mobile older household income plunges from over $137,000 a year to living on less than $35,000 a year in retirement income. The COVID-19 recession increases higher earners’ retirement insecurity by decreasing their financial preparedness for retirement. Their replacement rate falls 10 percentage points after the COVID-19 recession to 43% from 53% (see Figure 2). To put this decrease into perspective, a household earning $200,000 a year would lose $20,000 in annual retirement income, falling to $86,000 from $106,000 before the recession.


Policy Recommendations

Expand Social Security

Increasing the Social Security minimum benefit will reduce older-age poverty. Congress should immediately expand Social Security benefits by $200 per month and increase the Special Minimum Benefit up to 125% of poverty. This supports not only the most vulnerable — senior citizens, the disabled, people of color and those in densely populated areas — but also helps the overall economy by increasing aggregate demand.

Extend and Increase Unemployment Benefits

Older workers who are laid off experience longer spells of unemployment. Increased unemployment benefits —even more than the additional $600 a week—would bridge this gap, allowing older workers to preserve retirement assets and avoid claiming Social Security early, which lowers their monthly benefits.

Discourage Early Withdrawals

Congress should reinstate the 10% penalty fee for early withdrawals from tax-advantaged retirement accounts removed by recent passage of the CARES Act. Removing the fee encourages individuals to sacrifice their future needs for short-term spending. Rather, Congress should enact measures to ensure the income needs of those who lost jobs in the COVID-19 recession.

Create Guaranteed Retirement Accounts

Congress should ensure all workers have retirement plans that are protected against recessions through the creation of a public option retirement plan known as Guaranteed Retirement Accounts (GRAs). GRAs are universal, individual accounts funded by employer and employee contributions throughout a worker’s career along with a refundable tax credit (see Economic Policy Institute, American Federation of Teachers and Schwartz Center for Economic Policy Analysis 2019). The plan is portable as a worker changes jobs or if a worker loses their job. GRAs give workers access to a secure and accessible way to save for their retirement and supplement their Social Security benefits.

Unemployment Rates

The headline unemployment rate (U-3) for workers ages 55 and up was 13.6% for the month of April, up from the quarterly rate of 2.8%, which covers January to March but does not include the pandemic’s economic fallout. However, this aggregate number likely underrepresents older workers’ job loss because it does not include broader measures of unemployment, including the marginally attached, discouraged workers, involuntary part-time workers, and the involuntarily retired (those who say they want a job but have not looked in over a year). ReLab’s U-7 figure, normally reported here, includes these parameters to establish a full picture of the labor market for older workers. However, U-7 data that captures the pandemic’s effects are not yet available and will be reported in the next quarterly report.



Job Loss In The COVID-19 Recession

A worker is considered at risk of job loss if they cannot work from home and are not considered essential workers. If the overall unemployment rate reaches 25%, we project 10.5 million older workers will lose their jobs. For those in the lower half of the earnings distribution, 6.1 million older workers (29%) are projected to be laid off. For middle earners, 4.0 million (22%) could lose their jobs. And for high earners, making over the Social Security taxable maximum, 400,000 (15%) could lose their jobs. However, those who don’t lose their jobs but cannot work from home face increased risk of contracting the virus. Unfortunately, 40% of older workers do not have paid sick leave. Even among the 6.8 million older workers on the frontline, working jobs that are essential for the public health and safety, the share without sick leave is as high as 54% (Farmand and Ghilarducci 2020).



Labor Force Participation

Labor force participation among workers ages 55-64 fell to 63.9% in April 2020, down 1.5 percentage points from the first quarter of 2020. Labor force participation does not tend to move much, but last month’s decrease still represents 600,000 near-retirement workers leaving the labor force. Jobless older workers spent nearly twice as long looking for work on average compared to younger workers, and as the recession continues, many may decide to give up looking for work, and could become involuntarily retired (SCEPA 2016). This would be reflected in continued decreases in labor force participation in future months. 



Retirement Coverage

Many older workers who have not lost their jobs due to the COVID-19 recession are at risk of losing their retirement coverage, which was already at a historic low of only 41% in 2019. Further decreases in coverage due to employer cuts increases retirement insecurity for workers and weakens their bargaining position. 




1. Our poverty thresholds are twice the Federal Poverty Line ($25,520 for individuals and $34,480 for couples). Our thresholds are slightly lower than The Elder Economic Security Standard™ Index (Elder Index) threshold for elderly with good health. See Center for Social and Demographic Research on Aging (2017).

2. The appropriate target replacement rate is a matter of significant debate, and varies based on lifetime earnings, health, birth year, and household composition, but most experts advocate targeting a 60% to 100% replacement rate. While the recommended replacement rate for high earners is often between 60% to 70%, low earners require higher replacement rates, and those with incomes below or near poverty need replacement rates of more than 100%.

3. Low earners are more likely to claim benefits at 62, while middle and high earners typically postpone retirement and claim Social Security benefits at age 65 or older. While workers can increase their retirement income by claiming at older ages, many low earners—who are less likely to find adequate employment and have little savings—cannot afford to postpone claiming. In the COVID-19 recession, even more workers who lose their jobs will claim their Social Security benefits at 62. However, we do not include the possible change in claim age nor the earnings gradient in our model. See Ghilarducci and Webb (2018) for how retirement ages vary by class.

4. See footnote 3 about retirement age for different socioeconomic groups.

5. While the final effect of the crisis on asset values and the speed of recovery after the outbreak are unclear at this point, we assume a 20% drop in asset values in our projections.

6. Author's calculations, median savings of $4,800.

7. Most middle earners postpone claiming Social Security benefits to age 65. See footnote 3 for more detail.


Burkhalter, K. & Clingman, M. (2020). Scaled factors for hypothetical earnings examples under the 2020 Trustees Report assumptions. Social Security Administration Actuarial Note 2020-3.


Center for Social and Demographic Research on Aging, University of Massachusetts Boston and Gerontology Institute, University of Massachusetts Boston (2017). The National Elder Economic Security Standard™ Index: Methodology overview. Center for Social and Demographic Research on Aging Publications, 16.


Clingman, M., Burkhalter, K., and Chaplain, C. (2020). Replacement rates for hypothetical retired workers. Social Security Administration Actuarial Note 2020-9.


Economic Policy Institute, American Federation of Teachers and Schwartz Center for Economic Policy Analysis (2019). Breathe Easy—How Guaranteed Retirement Accounts could change your life. Retrieved from epi.org/16550.


Farmand, A., and Ghilarducci, T. (2020). Older Workers Are Underrepresented In “Safe” Jobs in the COVID-19 Recession. American Society on Aging. Retrieved from https://www.asaging.org/blog/older-workers-are-underrepresented-safe-jobs-covid-19-recession


Gascon, C. (2020). COVID-19: Which workers face the highest unemployment risk? St. Louis Federal Reserve. Retrieved from https://www.stlouisfed.org/on-the-economy/2020/march/covid-19-workers-highest-unemployment-risk


Ghilarducci, T., and Webb, A. (2018). The distribution of time in retirement: Evidence from the Health and Retirement Survey. Work, Aging and Retirement, 4(3), 251-261.


New York State (2020). Guidance for determining whether a business enterprise is subject to a workforce reduction under recent executive orders. Retrieved from https://esd.ny.gov/guidance-executive-order-2026.


Schwartz Center for Economic Policy Analysis (2016). Older Unemployed Workers Take Longer to Find Jobs than Younger Workers. Unemployment Report for Workers Over 55. Retrieved from https://www.economicpolicyresearch.org/jobs-report/december15-unemployment-report-for-workers-over-55


Schwartz Center for Economic Policy Analysis (SCEPA 2018). 52% of Older Workers Forced into Involuntary Retirement. Retrieved from https://www.economicpolicyresearch.org/jobs-report/september-2018-unemployment-report-for-workers-over-55 

 Technical Appendix

We project retirement incomes for older households by imputing career earnings and therefore Social Security benefits from current earnings and making generous assumptions for retirement savings accumulation (No Recession scenario). We then subject this projection to a simulated recession, where we project a 25% unemployment rate, concentrated in jobs identified as being at greatest risk of layoff in the current COVID-19 recession. We then assume all workers in the sample that are laid off in the recession find new jobs paying 20% less than current earnings.

This brief uses Wave 1 the 2014 Survey of Income and Program Participation (SIPP) and the supplemental questions in the Social Security Administration supplement. Workers’ individual retirement incomes are projected and summed into households. Retirement income is the sum of income from Social Security, defined benefit (DB) pensions, annuitized defined contribution (DC) savings, and annuitized wealth from other financial assets.

We construct three scenarios. In the “No-Recession” scenario, we assume full-length careers, to serve as a baseline. In the Recession-based scenarios. A total of 30 million new unemployment claims had been processed by the U.S. Department of Labor between March 15th and the time of writing, which constitutes approximately 18% of employed workers as of March. Many jobless workers are ineligible for unemployment benefits, and to reach an unemployment rate of 25% purely on the basis of those eligible for benefits, the current pace of claims would have to continue for just 2.5 weeks.

We identify workers as being at risk if they work in industries identified by the Federal Reserve of St. Louis as being at risk of layoff or working in an industry deemed by New York State (2020) to not be an essential occupation. “At risk” workers constitute 48% of the workers in our sample, meaning we would project 52% of them to become unemployed to reach a 25% overall unemployment rate.We randomly assign 52% of at-risk workers in our sample to be laid off. The figures presented here represent the average of 10 repetitions of this random sampling. In the Best Case scenario (presented in Figures 1 and 2), we assume laid off workers all find new jobs, and in the Worst Case scenario (presented only in the Appendix), we project none of them find new jobs and retire immediately. We caution the reader to not consider these scenarios as equally likely. Based on historical employment trends we expect a significant number to become involuntarily retired. However, we do not expect anywhere near 100% of job losers to become involuntarily retired, and therefore we believe our Best Case scenario is more realistic.

In our No Recession scenario, for households with two workers ages 44-54, for our age 62 scenario, we assume each spouse retires at age 62, project each spouse’s retirement income at age 62, and sum. For our age 65 scenario, if the younger worker is age 62 or younger at this point, we use their projected retirement income for age 62. If the younger worker is age 63-65, they receive their projected retirement income at that age. For spouses who have already retired, we take their current reported incomes from each income source. Only heads of household and their spouse (if any) are included as part of a household, and if there are multiple households living together they are treated as separate observations.

Because this survey only asks respondents to report their earnings from the most recent year, we must construct profiles of career earnings for each worker. The Social Security Administration constructs scaled earnings factors for ages 21-64, and we use these factors to construct age-earnings profiles for each worker. The 35 highest-earning years in these synthetic age-earnings profiles are then used to project Social Security income in retirement. In the Recession Best-Case scenario, we reduce future years’ earnings by 20 percent for those who lose their jobs and in the Recession Worst-Case scenario, we reduce future years’ earnings to zero.

We consider all DB plans from current and previous jobs to project DB pension income in retirement. For pensions from current jobs, in the No Recession scenario we assume the worker stays at their current job until retirement, and receives benefits equal to 1.5 percent of the average of their last five years of earnings at the job (using the synthetic age-earnings profiles) per year of job tenure. For the Recession scenarios, we assume the worker has enough tenure to be vested but leaves the job immediately. For pensions from past jobs, we assume the same accrual rate of 1.5 percent. For the purposes of determining earnings when transitioning out of past jobs, workers are assumed to have left past jobs at the same age and same nominal pay as their starting pay on their current job.

A worker’s DC savings is the sum of the balances in their savings in 401(k), 401(k)-equivalent accounts, and IRA savings, from current and past jobs. We project income post-retirement from retirement savings with generous assumptions. First, for years prior to 2020, workers earn returns equal to historical returns from the S&P 500 index, then lose 20% of their assets to the stock market crash in 2020, then earn 4 percent real return on investments net of fees from 2021 forward. Second, workers contribute 3.5 percent of earnings to their 401(k) with an employer match of 3.5 percent. Third, workers purchase an inflation-indexed annuity when they retire. Although people rarely purchase an inflation-indexed annuity, it provides a higher income than commonly used drawdown strategies and is the only financial product that provides an inflation-indexed lifetime income. Thus, the assumption yields a conservative estimate of the share of households financially unprepared for retirement. We assume April 2020 annuity rates. We make similar generous assumptions for income from other financial assets. A worker’s financial assets include the value of money market accounts, CDs, government securities, municipal and corporate bonds, stocks, and equity in annuities. We assume workers earn a 4 percent real return on their investments, and purchase an inflation-indexed annuity when they retire. For the Recession scenarios, we reduce future DC savings for laid-off workers to zero and assume workers draw down enough to replace half of their lost earnings in each scenario, then annuitize the remaining assets at ages 62 or 65.

We do not consider the impact of unemployment benefits on reducing retirement account withdrawals, since we assume immediate re-employment. We also do not consider any possible changes in contribution rates for those who do not lose their jobs, including workers whose employers stop matching contributions.






Suggested Citation: Papadopoulos, M., Fisher, B., Ghilarducci, T., and Radpour, S. Retirement Equity Lab. (2020). “Recession Increases Downward Mobility in Retirement: Middle Earners Hit From Both Sides.” Status of Older Workers Report Series. New York, NY. Schwartz Center for Economic Policy Analysis at The New School for Social Research.