Americans are concerned and even afraid for their retirement security. And the news headlines often don’t make them feel better. The latest is a claim from the National institute for Retirement Security that “A plurality of older Americans, 40.2 percent, only receive income from Social Security in retirement.” If true that’s very worrying. But does this frightening factoid hold up?
The National Institute for Retirement Security (NIRS) is the research and communications arm of the defined benefit pension industry. NIRS regularly produces reports on various aspects of retirement savings and incomes, and these reports usually generate favorable media coverage.
NIRS’s new report is no exception. “Retirement Security Is On A Treacherous Path, New Report Warns,” writes my fellow Forbes contributor Ted Knutson. The Washington Post’s Helaine Olen cited the report, arguing that because Social Security “is the only income source for 40 percent of retirees over the age of 60,” Democratic politicians who favor Social Security benefit cuts are vulnerable to President Trump, who opposes Social Security reductions. CNBC and others also repeated the claim that 40% of retirees have nothing except for Social Security.
But is the NIRS report correct? Do 4-in-10 retirees truly have no income other than Social Security?
The NIRS report’s data source is the Census Bureau’s Survey of Income and Program Participation (or, the SIPP). The SIPP surveys households by asking them a wide variety of questions, including the sources of their income. From the SIPP, NIRS declares that 40.2 percent of retirees receive all of their income from Social Security.
And yet, a 2017 study by researchers at the Social Security Administration, also using the SIPP, found that only 19.6% of Americans 65 and over received at least 90% of their total incomes from Social Security. That’s less than half the share of retirees than NIRS claims and SSA measures dependence using a lower bar – 90% of total income rather than NIRS’s 100%. Clearly, there’s a conflict. And, from a policy perspective, one-fifth of retirees being heavily dependent on Social Security’s isn’t a huge problem: the poorest fifth of workers are indeed quite poor, and Social Security was designed to provide a retirement benefit for workers who can’t easily save on their own.
Moreover, a second 2017 study, from two Census Bureau economists, analyzed retirement incomes using IRS tax records, which are more accurate than households’ responses to a survey. The Census Bureau study found that only 12% of Americans aged 65+ received 90% or more of their income from Social Security. Again, it’s not clear how that is compatible with NIRS’s claim that over 40% of retirees receive all their income from Social Security.
The Census Bureau study shows that even if you combine Social Security benefits with Supplemental Security Income (SSI), a means-tested welfare benefit paid to very low-income retirees, retiree households in the bottom 40% of the income distribution – the ones the NIRS report would have you think are receiving pretty much all of their incomes from Social Security – received only 79% of their total incomes in combined Social Security and SSI benefits. As we move up the income distribution, dependence on Social Security/SSI declines further and reliance on other sources of income increases.
Based on these findings alone, I think it’s fair to give NIRS’s claim that 40% of retirees receive all their income from Social Security a failing grade on the Biggs Factcheck.
The more interesting question is why: How did NIRS produce such a figure? Without recreating NIRS’s data from scratch it’s hard to say.
One difference between NIRS’s calculations and those from the SSA, the Census Bureau and others is how NIRS defines “retirees.” Most studies of retirees’ dependence on Social Security look at all Americans aged 65 and over. Instead, NIRS looked at Americans age 60 and over who work less than 30 hours per week. Obviously, anyone who is working has income other than Social Security, so NIRS’s definition of retirees does reduce the number of retirees with non-Social Security income. But from my quick look at Current Population Survey data for 2013, only about eight percent of American aged 60 and over worked more than 30 hours per week. So I doubt that’s the driving factor.
Another factor may be NIRS’s reliance on SIPP’s pensions and retirement supplement, where survey respondents report whether they receive various types of retirement income. In theory this should increase the accuracy of NIRS’s report, but other studies have found that retirees aren’t great at reporting their difference sources of income. For instance, they might claim not to be receiving a pension benefit even if tax data show they are. The Census Bureau study found that 28% of Americans 65 and over claimed not to receive income from a private retirement plan even when tax records showed they did. Only four percent of retirees said they have private retirement plan benefits when they didn’t. So errors from household surveys tend to make things look worse than they really are. That’s a big reason studies have increasingly relied on administrative data from the IRS or the SSA rather than households surveys.
But we can still safely conclude that NIRS’s claim that “A plurality of older Americans, 40.2 percent, only receive income from Social Security in retirement” is incorrect. Analysis by government agencies conclude that only about half that amount receive even 90% of their incomes from Social Security.
