Question Err?

“Presumably, if workers earned income in a retirement account, it is safe to assume that they had a retirement account…”.

Ya think?

That somewhat self-evident statement is drawn from a recent Issue Brief by the non-partisan Employee Benefit Institute (EBRI). That it was necessary is a cautionary tale about the blind reliance on data, even from a credible source, that looks suspicious.

It relates to the Current Population Survey (CPS), Annual Social and Economic Supplement (fielded in March of each year) to the CPS, conducted by the U.S. Census Bureau – a report that had long been one of the most cited sources[i] of income data for those whose ages are associated with being retired. 

‘Bold’ Move

The problem appears to have its roots in a well-intentioned attempt to provide a more accurate read on income from DC plans. Responding to research that indicated that the CPS misclassified and generally underreported income, particularly pension income, the Census Bureau in 2014 conducted a redesign of the CPS questionnaire that altered and added to questions on income in order to better capture income from pensions.

To do so, the authors of this survey added (IN BIG BOLD LETTERS) an instruction that respondents NOT INCLUDE DISTRIBUTIONS OR WITHDRAWALS FROM IRAS, 401(k)s, or SIMILAR ACCOUNTS. This “clarity” does, in fact, seem to have produced a more accurate read on that specific question. The reporting of pension income that year, and in subsequent iterations, has risen.

However, beginning with that version of the CPS, there was a commensurate sharp drop in both the overall percentage of workers participating in a retirement plan, a decline in the number of workers participating, and declines in participation among those most likely to participate. This result, by the way, was counter to other reputable data sources, including the upward trend in the number of active participants in private-sector plans according to tabulations of Form 5500 filings by the Employee Benefits Security Administration, findings on retirement plan participation from the Bureau of Labor Statistics’ National Compensation Survey (BLS-NCS). 

Fall ‘Call’

How much did it drop? The EBRI report explains that for full-time, full-year wage and salary workers ages 21-64, the percentage participating in 2013 under the traditional questionnaire was 54.5% vs. 49.3% under the redesigned questionnaire. This number fell to 39.9% in 2018 despite a slight uptick to 41.4% in 2017. Moreover, the number of full-time, full-year wage and salary workers ages 21-64 estimated to be participating in an employment-based retirement plan decreased from 51.4 million in 2013 (under the traditional design) to 41.0 million in 2016. By the way, the number of active participants increased from 89.9 million in 2014 to 94.6 million in 2017 according to 5500 data.

So why did the question regarding pension income translate into such enormous declines in participation? Well, you have to engage in a bit of speculation, but the trend and the timing suggest that individuals responding to the CPS, told to exclude 401(k)s, 403(b)s, IRAs, etc. on the question regarding pension income basically applied that direction to questions regarding their participation in a plan. 

Gap ‘Flap’?

The good news is that this aberrant result was caught almost immediately by EBRI, and later by groups such as the Investment Company Institute. And, when a subsequent version came out with the same aberrant results, those groups once again held that up to scrutiny. And yet, that CPS data was still “out there,” and for some constituted the “official” version of the state of retirement plan participation in the United States.

That said, the newest iteration of the CPS included a new question that might be a step along the way to a more accurate assessment. Specifically, EBRI researcher Craig Copeland notes that in the 2019 dataset, the CPS added variables relating to income earned in a retirement account – and that addition, coupled with the assumption that began this post, provides some hope for an accurate data read.

That additional question simply posits whether the respondent received interest income, and from what source(s) – including 401k, 403b, among other retirement accounts. Copeland has taken the number of workers who responded that they received income from one of the employment-based retirement accounts and added that to the number of workers who reported having a pension plan under the traditional questions (accounting for those responding yes to both) to provide an adjusted overall estimate of employment-based retirement plan participation.

What’s the Impact?

Recall for a moment the data on fulltime, full-year wage and salary workers ages 21-64, that had indicated a 54.5% participation in 2013, only to drop below 40% in 2018 with the new questionnaire. However, taking into account the information from the new question, Copeland finds that the adjusted percentage participating jumps to 59.0% for this group in 2018. Similar, positive results are found for the other categories as well – results, it should be noted, that are much more in line with other data sources.

Now, despite those positive results, Copeland cautions since this is just one year of data, “a trend is obviously indeterminable.”

That said, I can’t help but be reminded about that old story about the optimistic child who awakens on Christmas morning and finds a heap of horse manure under the tree, rather than the anticipated – and still responds, “With all this manure, there must be a pony somewhere!”

Seems as though Dr. Copeland has found a pony in this data, after all.

[i]See Data ‘Minding. Compounding the issues, for participation data, the NIRS draws on the Current Population Survey (CPS), even though the report’s own footnotes acknowledge that “the 2014 redesign of CPS produced much lower participation rates for working Americans in years after 2014 (participation in 2014 was at 40.1% but at 31.7% in 2017), which has not been fully explained.” They aren’t the only ones to take note of this aberration (see CPS Needs a New GPS and Commonly Cited Participation Gauge Misses the Mark, Study Says), but decide, for reasons not fully explained, to rely on the data there anyway. Indeed, a June 2018 report on the impact of the changes in the CPS by the nonpartisan Employee Benefits Research Institute (EBRI) cautions that “the estimates from the most recent surveys could easily be misconstrued as erosions in coverage, as opposed to an issue with the design of the survey.”