Unconventional Wisdom on Retirement Preparedness

How well are Americans planning and saving for retirement? This is an important question to answer—but also vexingly difficult. Researchers have come to different conclusions. Much of the academic research and ICI’s own analysis of the data indicate that, contrary to conventional wisdom, most Americans are properly preparing for retirement.

To better inform the public dialogue, ICI hosted a retirement summit last spring focused on the issue of preparedness. The summit featured some of the nation’s top retirement researchers, with the afternoon spent discussing four studies that reached widely different conclusions on how well Americans are planning and saving for their old age. The amount of resources that households had accumulated for retirement was not hotly debated, but the amount of resources needed in retirement was.

The results of these four studies are summarized in the graphic below, with the green portion of the bars representing the estimated percentage of Americans prepared for retirement—according to the standard set by that study—and red representing the percentage who fall short of that study’s standard. (The results of all of the studies are sensitive to assumptions. The most commonly cited estimates are presented.)

Four Estimates of Retirement Preparedness in America

Note: The results of all four studies are sensitive to assumptions. The chart shows the headline results for the first three studies and representative results from a series of papers for the fourth.

Source: Based on presentation by Steve Utkus, “Four Views of Retirement Preparation,” available here. 

Readers may be most familiar with the two studies at the top of the chart. The National Retirement Risk Index (NRRI), published by the Center for Retirement Research (CRR) at Boston College, estimated that 48 percent of American households are prepared, with the remainder “at risk” of being unable to maintain their pre-retirement standard of living in retirement. The Retirement Readiness Ratings of the Employee Benefit Research Institute (EBRI), using a standard of having enough retirement income to cover average expenses and uninsured healthcare costs, found 57 percent of the population prepared, with the remainder at risk.

The two studies use vastly different methods to arrive at their estimates, and both are serious and technically rigorous studies. That said, we believe that both overstate the amount of resources households need in retirement.

Unrealistic Assumptions

The NRRI’s standard for adequate retirement resources is based on a replacement rate: that is, to maintain their standard of living, workers must be on track to have enough resources to generate a fixed percentage of pre-retirement income throughout retirement. Although the study is arguably the most rigorous of its type, replacement rate studies fail to take into account the realities of household behavior over a lifetime and end up setting the bar too high. (For a more thorough discussion of the shortcomings of replacement rate studies, see this recent National Affairs article by Andrew Biggs and Sylvester Schieber.)

There are at least three reasons why adequacy standards in the typical replacement rate study are too high:

• Assumptions about the pattern of savings over workers’ life cycles raise the standard for younger workers. The typical assumption in these analyses is that workers save steadily for retirement from the start of their career until retirement. In fact, most don’t do this. Early in their career, workers typically have lower earnings and are focused on other savings priorities: paying for education, purchasing a home, starting a family, and building up a “rainy day” fund. Later in their working career, when earnings are higher and those savings needs are met—and especially when child-rearing expenses decline—workers turn their attention to retirement savings. This pattern makes it difficult to assess retirement preparedness for younger workers who are not in or near retirement.

When studies use adequacy standards that fail to account for this life-cycle savings pattern, the studies misjudge how well workers—particularly younger workers—are preparing for retirement. Take, for example, a household that reaches retirement with adequate resources, but which achieved that adequacy by saving in an uneven life-cycle pattern. Relative to a smooth savings path, the household would have been judged to be falling short at any point before retirement—at age 30, age 40, age 50—despite being on track for a secure retirement throughout.

• The assumption that household expenses don’t change when households transition from work to retirement raises the standard. Replacement rate studies do not adequately account for the reduction in household expenses when work ends and retirement begins. The NRRI study does better than most in this respect in that it reduces its estimates for income and payroll taxes—which typically decline in retirement—when calculating target replacement rates. It also attempts to account for the benefits of homeownership (homeowners do not need to generate as much monthly income as renters to cover housing expenses).

However, the evidence suggests that household expenses decline for other reasons that the NRRI study ignores. Work-related expenses—such as commuting costs, clothing expenditures, and fast-food purchases—decline at retirement. Perhaps most importantly, retiree households typically are no longer incurring the expense of raising children. A couple who have raised three children can maintain their accustomed standard of living on less money—and a lower replacement rate—than can a childless couple who have supported only themselves on the same income. Replacement rate studies ignore that difference.

• The assumption that household expenses remain flat throughout retirement raises the standard. Even when replacement rate studies account for some decline in expenses when a household retires, they often assume that household spending stays constant throughout retirement. This is in contrast both to empirical analysis showing that household spending on average continues to fall during retirement and to economic models that predict households rationally choose to consume less as they age.

The second study—EBRI’s Retirement Readiness Ratings—defines adequacy as having enough resources to cover average expenses and uninsured health costs (including long-term care expenses). The EBRI model differs considerably from the approach used in the other three studies and is too complex to summarize succinctly. Suffice to say, the model is very rich, but its results are sensitive to changes in the underlying assumptions—as the authors themselves point out. For example, they found that the headline statistic of 57 percent of the population with adequate resources increased to more than 80 percent when they used alternative assumptions about health expenses.

Assumptions Consistent with Observed Behavior

Readers may be less familiar with the next two studies in the graphic above. These studies do a better job of accounting for the way that households typically accumulate wealth and change expenditures over the life cycle. Thanks to this grounding in real behavior, they reach much more optimistic conclusions about Americans’ readiness for retirement.

RAND economists Michael Hurd and Susann Rohwedder estimated that 71 percent of households have adequate retirement resources. Their study focused on households just past normal retirement age (ages 66 to 69), enabling the authors to focus on the resources that retirees actually have and ignore whether the retirees had saved steadily or unevenly while working. Accounting for taxes and out-of-pocket health spending, the study judges a household as properly prepared if it can maintain its consumption to the same extent as previous generations of retirees (with expenditures slowly declining with age).

The fourth bar in the graphic above summarizes a series of studies by the University of Wisconsin’s John Karl Scholz and Ananth Seshadri, who measured the share of U.S. households on track to maintain their standard of living in retirement. Their studies use detailed data on wealth and lifetime earnings for a representative sample of households approaching or in retirement (the age range differed by study, but all were age 51 or older). For each household, they calculated the optimal amount of wealth using a lifecycle model of consumption—based on the model’s prediction of how much the household should have saved—and compared that to the household’s actual wealth. Importantly, they accounted for the effect of children on household expenses and, consistent with observed behavior, predicted that consumption declines with age in retirement.

Their original study, based on data for households approaching retirement age in 1992, found that 84 percent of households had wealth equal to or greater than their optimal targets. When the researchers updated their original study using 2004 data, they found an even higher percentage at or above optimal wealth. They then reexamined the 2004 data using even more stringent adequacy standards and found that, even after ignoring half of homeowners’ housing wealth, the share with optimal or greater wealth only fell to 74 percent.

The Danger of Conventional Wisdom

The RAND and Wisconsin studies, in addition to ICI’s own analysis, are at odds with the conventional wisdom that Americans aren’t preparing for retirement. The danger of accepting the conventional wisdom is two-fold. First, the notion of a widespread shortfall deflects attention from those groups actually at risk in retirement—such as those retiring early because of poor health and those with limited work histories. Second, it obscures the fact that the U.S. retirement system works well for the majority of households. In turn, the mistaken conventional wisdom can encourage misguided policy proposals that actually undermine the current retirement system without addressing real problems.

We believe that the most prudent path for policymakers is to build on the success of the U.S. retirement system and to focus on helping those who appear to be most at risk.