Tweet This
Note: See updated figures at close of article.
Last year, Syl Schieber and I wrote in the Wall Street Journal about how popular statistics on retirement income dramatically understate the amounts that retirees collect from IRAs and 401(k)s. These statistics show that Americans are highly – and increasingly – dependent on Social Security for their retirement income, causing many to conclude that Americans face a “retirement crisis” and that personal saving plans such as IRAs and 401(k)s have failed. Yet the Census Bureau’s Current Population Survey, which is used by the Social Security Administration in its popular Income of the Aged series, defines as “income” only regular payments such as the monthly checks retirees receive from Social Security or traditional “defined benefit” pensions. Irregular payments, such as as-needed withdrawals from “defined contribution” 401(k) plans and Individual Retirement Accounts, aren’t counted as income.
Our op-ed, which drew on a more detailed paper co-authored by Schieber and Billie Jean Miller, compared the income that the CPS claims retirees receive from IRAs and 401(k)s to the income that retirees report to the IRS. Since you are required to pay taxes on IRAs and 401(k)s when you collect benefits, IRS data should do a pretty good job of catching these income sources. IRS data isn’t perfect: low-income retirees may not need to file returns, and for confidentiality reasons IRS data isn’t as detailed as other data sources. But these figures showed that the CPS was capturing far less retirement income than IRS data. We wrote:
“For 2008, the CPS reported $5.6 billion in individual IRA income. Retirees themselves reported $111 billion in IRA income to the Internal Revenue Service. The CPS suggests that in 2008 households receiving Social Security benefits collected $222 billion in pensions or annuity income. But federal tax filings for 2008 show that these same households received $457 billion of pension or annuity income.”
So if you’re relying on the CPS or SSA publications that rely on CPS statistics, you’ll probably conclude that a) Americans don’t have nearly enough retirement income, and b) that IRAs and 401(k)s aren’t particularly good retirement saving vehicles. The problem is that you’re drawing those conclusions by ignoring the income that American retirees derive from IRA and 401(k) plans. Not a good research strategy.
Analysts at the Social Security Administration have been well aware of these income measurement issues. Indeed, SSA’s policy shop has been a leader in drawing attention to them. Recently, the Census Bureau introduced new questions to the Current Population Survey designed to better measure asset income, including retirement income from IRAs and 401(k)s. The results of the updated CPS are reviewed in a new article by Craig Copeland of the Employee Benefit Research Institute.
For Americans age 65+ in 2013, the CPS using the old questions picked up $254 billion in combined income from traditional pensions (private sector, state and federal), 401(k)s, IRAs, and annuities. Using the new questions, the CPS captured $324 billion in income from those sources, a 28 percent increase. Even then, however, the increase in overall retirement incomes due to more accurately counting IRA and 401(k) benefits isn’t massive, because even the revised CPS doesn’t find large amounts of retirement income coming from IRAs and 401(k)s.
But there’s reason to believe the CPS is still understating retirement, even with new-and-improved questions. One reason is that a different data source, the Survey of Income and Program Participation, has long asked about retirement income from assets. And, according to SSA analysts, the SIPP’s aggregates still fall short: “the SIPP asks about distributions from retirement plans, but comparisons with IRS data indicate that the SIPP greatly underestimates the amounts of such distributions.” So it could be that, even with better questions, surveys aren’t as good as administrative data at picking up the as-needed withdrawal approach that most retirees use with IRAs and 401(k)s.
To check, I did a quick comparison of the new CPS figures with IRS data on income from retirement plans. The IRS hasn’t released figures for 2013, so I used 2012 data adjusted upward by the growth rate from 2011-2012. These adjusted figures show, for individuals aged 65 and over, $168 billion in taxable IRA distributions and $423 billion in taxable income from pensions and annuities, for a total of $590 billion. The aggregated IRS data don’t break out traditional pensions from 401(k)s, so we compare total pension and annuity income. By restricting to taxable income for older households, I’m hoping to catch actual income that retirees spend rather than rollovers from one retirement plan to another.
This combined $590 billion in income from retirement plans far exceeds the $324 billion reported by the revised CPS. In other words, it’s likely that retirees are still doing substantially better than even the new CPS reports. And that's good news for retirees.
Does this mean that everything is hunky-dory? Not at all. Most truly low-income retirees don’t have IRAs and 401(k)s. Survey data can't under-report what they don't have. But the “retirement crisis” theme is that it’s not just the poor who aren’t prepared, but middle class and even upper income people as well. These are the folks who have defined contribution pensions. A more accurate picture of their retirement income is likely to be a more encouraging picture as well.
The figures above are from IRS totals, not the more detailed micro-data that Schieber and Miller used for their article. And there are definitional issues that mean that things don’t always match up precisely. But these figures lead me to conclude that the Current Population Survey, and the SSA publications that rely on it, still aren’t catching all the income that American retirees receive from IRAs and 401(k). If you want to know how well prepared Americans are for retirement and what we should do to improve retirement saving, these are important questions to answer.
Update: Since I wrote this piece, the official IRS Statistics of Income data for 2013 have been released, meaning that I no longer need to rely upon estimates based upon 2012 IRS figures. The official IRS data bear out my argument: while the revised CPS questions on IRA, pension and annuity income do increase the amounts reported by retirees, from $254 billion to $324 billion, the IRS reports $563 billion in such income to individuals ages 65 and over. In other words, even the revised Current Population Survey undercounts retirement plan income by 42 percent. Relying on the CPS to judge Americans' retirement income adequacy could lead to an overly pessimistic picture of where we currently stand.
