Many retirement research analysts and policymakers believe that most American workers are not on a viable path to a financially secure retirement. Fifty-three percent of working-age households are at risk of having insufficient retirement resources, according to one analysis,1 and 84% of workers are behind savings targets, according to another.2 To maintain living standards into old age, workers are advised to accumulate eight, 11 or even 20 times their annual earnings in personal financial resources for retirement.3 The task looms so large that one researcher concluded that saving adequately through the modern 401(k) “do-it-yourself pension system” is beyond the capability of most workers.4
This article describes approaches to defining retirement income goals and some of the measurement flaws and discrepancies behind the dire predictions for American workers’ retirement. Carefully considering workers’ income, consumption and savings patterns over their life cycle reveals undue overestimates of earnings to be replaced in retirement and misperceptions about workers’ own responsibility for securing their retirement prospects. We believe that American workers are generally better equipped for retirement than depicted in some studies. We stop short of suggesting a universal model or target for retirement savings because the one-size-fit-all approach would be out of touch with the varying household situations.
Measuring savings constraints and capacity over the life cycle
Assessments of workers’ retirement savings are usually built off a life-cycle model for distributing career earnings evenly across the adult lifespan, with consumption during retirement financed from savings over the career. Consumable income is what remains after netting out expenses related to working and paying taxes.
To estimate consumption levels both before and after retirement, analysts often use survey data on income, consumer expenses and spending behavior. These amounts are then used to estimate target income replacement rates: the percentage of preretirement income required to maintain working standards of living during retirement. In regard to health care for retirees in particular, some analysts use survey data and then add estimated health insurance premiums and out-of-pocket health expenses, plus estimated probabilities of long-term care and related expenses at advanced ages.5
Analysts working with administrative data that reflect contributions to employer-sponsored plans often project retirement accumulations based on current savings rates and annual rate-of-return assumptions. Extrapolating from a young worker’s financial situation to retirement age, however, is shaky. Workers save at different rates over their lifetimes and accumulate capital in various ways. They buy homes and start businesses, and many also save in individual retirement accounts or other employer-sponsored retirement plans.
Administrative databases do not include all these assets and so paint an incomplete picture of workers’ savings. Survey data might provide more complete financial information, but self-reports often have lower accuracy.
Most retirement income models do not account for the presence of children in the household. Studies using lifetime earnings histories have found that parents generally have less net worth than those without children because of higher consumption levels when children are present.6 Consumption levels tend to decline when children leave the home, and several studies have found that controlling for the number and sometimes the ages of children is important in explaining savings rates and accumulated savings.7 Some researchers have suggested that adult children are living with their parents longer than they used to, but a review of Current Population Surveys shows that, while the prevalence of adult children living at home has risen slightly in recent years, it has generally been declining over the past decades.
Home equity is another important factor often overlooked in setting income targets. According to the 2004 Survey of Consumer Finances, 81% of households headed by 65- to 74-year-olds owned a home and 60% were mortgage free.8 Thus, many retired homeowners should need that much less income to maintain a comparable standard of living in retirement.
Setting replacement rates (more) scientifically
Unsurprisingly, estimated replacement rates and retirement savings targets vary widely. Fidelity Investments suggests a savings target of eight times annual earnings at retirement, which, with Social Security, should provide retirees with an income of 85% of preretirement earnings. Aon Hewitt holds that average-earning full-career workers should amass assets worth roughly 11 times their final salary to replace 85% of preretirement income.
A study by Scholz and Seshadri used two measures of preretirement earnings for retired households: the sum of lifetime earnings indexed by the cost of living and average real earnings in the five years before retirement. Because reported wages often decline as workers approach retirement and scale back work levels, they measured preretirement standards of living as average earnings in the five to nine years before retirement.9 At every earnings level, their estimated replacement targets were in the 46%-to-76% range for different earnings levels and marital status, significantly lower than the 85% targets set by Aon Hewitt and Fidelity.
Target replacement rates are also much lower than 85% under the model developed by the Center for Retirement Research at Boston College. The Center’s measure of preretirement income includes earnings, returns on retirement plan assets, income on other financial assets net of non-mortgage debt and “imputed rent from housing (net of interest paid on mortgage debt).”10 The Center uses the average wage index series to determine preretirement earnings.
Determining whether workers are on track
Are workers on track to achieve adequate incomes for their retirement years? Nari Rhee used Fidelity’s “Age-Based Savings Guidelines” to address this question in her National Institute on Retirement Security report. Fidelity intended these guidelines as “rules of thumb,” but Rhee applied them rigidly to assess whether workers are saving adequately. Rhee’s assessment held that workers should accumulate savings equivalent to their salary by age 35, three times salary by 45, five times salary by 55 and eight times final salary in the year before retirement. Rhee ignored Fidelity’s stipulation that its savings goals were intended for average earners. She failed to consider that Social Security’s relatively high benefits for lower earners would reduce the amount they needed to save, or that higher earners would need to save more under the model.
Rhee also ignored the reality that there is more than one path to successful career savings. Her analysis assumed that workers would save 6% of pay at age 25, increase their savings rate by 1 percentage point of pay each year up to 12%, and then save at that rate until retirement.
What if, instead of following that arbitrary timetable, a worker contributed 6% per year from age 25 to age 40, after which he or she increased the contribution rate by 1 percentage point per year up to 20%? The worker would be behind the Fidelity path by 21% at age 30, by 34% at age 40, by 23% at age 50 and exactly on target by retirement age. After getting early life debts and start-up expenses under control, many workers can ramp up their contributions during peak earning years and as children are launched. Yet at every point but the end point, these workers would be deemed at risk of an impoverished retirement.
How much do workers really need to save?
The models used to estimate whether workers are saving enough for retirement either explicitly or implicitly build off Social Security. The Social Security Administration (SSA) regularly publishes its estimated Social Security replacement rates for hypothetical workers. Their actuaries use the national average wage index to adjust the workers’ earnings to age-64 levels and then calculate the average of the highest 35 years of indexed earnings (“SSA’s method,” hereafter) as the denominator for income replacement rates.11 Most retirement studies use a different measure of average earnings.
We estimated Social Security benefits for five hypothetical workers at various earnings levels. These estimates are based on the SSA’s method (indexed to age 64), average indexed monthly earnings (AIME) officially used to calculate the Social Security benefit (indexed to age 60), the CPI-W index (indexed to age 64) and average earnings from the worker’s last five working years (nominal).
According to SSA’s actuaries, Social Security benefits would replace 41.5% of the medium earner’s preretirement earnings, as shown in Figure 1. Using the high 35-year average of price-indexed earnings would push the Social Security replacement rate at least five percentage points higher, and using average earnings over the final five years would boost the replacement rate by 15 percentage points for the medium earner and by 12 percentage points for the high earner.
Figure 1. Estimated Social Security benefits and replacement rates for hypothetical workers born in 1949, retiring at age 65 in 2014
Click Image to Enlarge
Source: Developed by the authors
According to our calculations, medium earners would not need eight or 11 years’ worth of earnings to replace 65%, 75% or even 85% of their preretirement income by most measures (Figure 2). Based on price-indexed earnings, the worker would need about six years’ worth of earnings to meet the 75% target and four years for 65%. Many workers who saved eight times earnings would have significantly higher spendable income in retirement than they did while working. Social Security would enable very low earners to maintain their preretirement income levels without any supplemental savings — indicated by dashes in the table. To be clear, there are substantial retirement savings needs for most workers regardless of preretirement income measures or replacement targets used. However, the results indicate that a lifetime savings target of eight to 11 times earnings at retirement age, as suggested by some analyses, would overstate the savings needs of the majority of workers. Why should people scrimp on substandard consumption levels while they’re working to finance a higher standard of living in retirement?
Figure 2. Supplemental savings requirements as a multiple of the specified earnings base for hypothetical workers born in 1949, retiring at age 65 in 2014
Click Image to Enlarge
Source: Developed by the authors
Assessing policy proposals
The perceived retirement savings “crisis” has prompted several proposals to ameliorate the problem. One proposal would curtail tax preferences for retirement savings plans and use the money to finance a supplemental flat Social Security benefit, which would bring combined average Social Security benefits up to 60% of a medium earner’s preretirement earnings.12
Another proposal would replace current tax preferences for retirement savings with a mandated 5% annual contribution on earnings up to the Social Security tax base that would accumulate in accounts managed by Social Security. These accounts would provide guaranteed returns, and low earners would receive tax credits. According to the proposal, the new payouts combined with existing Social Security benefits would boost replacement rates to 89% for low earners, 71% for average earners and 61% for high earners.13
These proposals would adopt income security policies that significantly reward some people much more when they are not working than when they are, as indicated in Figure 3. Creating a savings-based earnings replacement system where benefits exceed preretirement income levels makes little sense in the context of income smoothing across the life cycle, which is part and parcel of what the retirement income security system is intended to achieve.
Figure 3. Estimated replacement rates under proposals for hypothetical workers born in 1949, retiring at age 65 in 2014
Click Image to Enlarge
Source: Developed by the authors
Most of the recent assessments suggesting that the majority of American workers are insufficiently prepared for retirement are based on models that fail to reflect patterns of income, consumption and savings that vary over workers’ lives. They extrapolate younger workers’ observed savings behavior into the future, ignoring their capacity to catch up after children leave home and the mortgage is paid off. Many measures of preretirement income and consumption are overstated owing to inappropriate indexing, which leads to overestimates of earnings replacement targets and underestimates of the income replacement capacity of Social Security.
Although clearly some workers are not saving enough for a comfortable retirement, the situation is less dire than many studies have suggested. And poorly conceived standards of how much workers will need as they approach retirement, naive models of retirement savings behavior and underestimates of existing savings programs cannot help us discern how many workers are at risk, who they are and how best to help them.
This summary discussion is based on Gaobo Pang and Sylvester J. Schieber’s paper, “American Workers’ Retirement Income Security Prospects: A Critique of Recent Assessments,” mimeo, 2014.