At Last, Some Action On Misleading Retirement Income Statistics | #retirement | #elderly | #seniors
Retirement income security is one of Americans’ biggest worries. And facts and figures on retirement incomes are the way that citizens, activists and lawmakers convey those concerns. But what if some of the most prominent figures aren’t actually facts? New federal government research shows that many common factoids on retirement income adequacy are just plain wrong. And federal agencies are finally starting to recognize the problem.
Back in 2012, I co-authored a Wall Street Journal op-ed with retirement expert Sylvester Schieber arguing that the federal government’s main source of data on retirees’ incomes – the Current Population Survey (CPS) – severely undercounts the benefits retirees draw from private retirement plans, such as traditional “defined benefit” pensions and retirement accounts like IRAs and 401(k)s.
Our evidence was simple: in the CPS, a household survey in which a sample of Americans are asked to report their various sources of income, retirees in 2008 reported receiving less than $250 billion in income from pensions and retirement accounts. The Internal Revenue Service, by contrast, reported that retirees received over $450 in income from private retirement plans, based on data the plans themselves are required to submit. Clearly, a lot of retirement income wasn’t being counted.
The reasons the CPS understates retirement plan benefits aren’t totally clear. One problem is the way the CPS defines “income,” which includes only regular payments, such as monthly pension benefits. This definition excludes as-needed withdrawals from IRAs or 401(k)s, which are increasingly common. But the CPS also understates benefits even from traditional pensions. The best explanation here is that survey respondents just answer incorrectly.
Either way, comparisons to IRS data show that the government’s most prominent dataset is lowballing retirement incomes. The upshot is that U.S. retirees’ incomes are higher, and their poverty rates and dependence on Social Security are lower, than is commonly understood.
For instance, based on response to the Current Population Survey, you might read that 9.5% of U.S. seniors had incomes below the poverty threshold in 2019. Or that retiree households had median incomes in just 2016 of $39,823. Or that in 2014, one-third of Americans 65 and over received nearly all of their incomes from Social Security benefits. All these factoids lead to the conclusion that the U.S. retirement system isn’t working very well. And all of them are false.
Today we have a much better understanding of how official statistics understate retirees’ incomes, thanks to two recent studies from the Census Bureau and the Social Security Administration.
The first paper, by Census Bureau economist Adam Bee and former Census economist Josh Mitchell, now at Welch Consulting, matched Current Population Survey responses to data from the Internal Revenue Service. This allowed Bee and Mitchell to compare the income that CPS respondents reported in the survey to the actual amounts the IRS knew they had.
Unfortunately, the Bee and Mitchell paper isn’t yet publicly available, but the pair did present their results at the January 2021 conference of the Allied Social Sciences Association. And their data showed a much more encouraging picture retirement income adequacy.
For instance, the CPS data Bee and Mitchell accessed showed that in 2016, the median senior household had a total income of $40,820. But IRS data show the median household aged 65+ had a true income of $52,180. It’s hard to overstate how big that 28% difference is. For context, average Social Security benefits grow about one percent faster than inflation each year, and it’s not unreasonmable to assume that other sources of retirement income would keep pace. So Bee and Mitchell “found” about a quarter-century’s worth of retirement income growth, simply by getting the data right.
With higher retirement incomes comes lower old-age poverty. Instead of 8.8% of Americans 65+ having incomes below the poverty threshold in 2016, as shown in CPS data, the Census Bureau study showed the true poverty rate was just 6.3%.
The IRS data also showed remarkable growth in retirement incomes over time. From 1990 to 2016, the typical retiree household’s income rose by 38% above inflation, while the share of retirees in poverty fell by one-third. These are signs of a retirement system that’s working.
A second study, released in January 2021 the Social Security Administration, confirms these results. The study’s authors, Irena Dushi and Brad Trenkamp, also compare the Current Population Survey to IRS administrative data. And they again find that having accurate data tells a more encouraging story about retiree incomes.
For instance, in the CPS sample only 47% of Americans 65+ in 2015 reported receiving private pension benefits. The IRS figures show that in reality 69% of all retirees received private plan benefits, a huge difference.
The SSA study salso shows that the well-worn factoid that over one-third of retirees receive 90% or more of their incomes from Social Security is wrong, for two reasons. First, the one-third of retirees reference is really a reference to one-third of retiree households. But married retirees are less dependent on Social Security than are single retirees, which means that, even using the CPS data, only 24% of actual seniors relied so heavily on Social Security. But then when SSA used IRS data to account for the CPS’s underreporting of retirement incomes, less than 14% of seniors received 90% or more of their incomes from Social Security. We go from one-third of retirees heavily reliant on Social Security to less than 14% just by getting the numbers right.
All of this led to a welcome outcome: the Social Security Administration’s Fast Facts About Social Security, a common source for the retiree income statistics cited above, now has suspended publication of retiree income statistics based on the Current Population Survey, citing concerns regarding the reliability of those data. This may anger activists, who cite such figures to claim that retirees are suffering, private savings are inadequate and, increasingly, that we should not merely maintain but expand Social Security benefits.
But we will soon face an interesting event: will the SSA, which increasingly will be staffed by appointees of President Biden, succumb to activist pressure and revert to using retiree income statistics that we now know for certain are mistaken? Or will the Biden appointees stand up for quality public policy research, even if it may undercut the administration’s political objectives?