Questions are focused on income that comes from annuity payments, like those from a defined benefit plan.

That means significantly less data is recorded on the income retirees derive from non-annuitized sources, like distributions from 401(k) plans and IRAs.

Aware of the inaccuracies, the Census Bureau revised questions and asked them of a portion of respondents in 2014. The EBRI’s study examined the differences in respondents’ answers to the old and new sets of income questions.

“The new measure of income in the CPS finds significantly more income that is from IRAs and 401(k)-type retirement plans than what has been reported under the old measure of income in CPS,” said Craig Copeland, senior research associate at EBRI.

Compared to the amount of total income accounted for in the traditional set of questions, the new, revised questions resulted in an additional 9.1 percent of total annual income, or an increase of $133 billion.

When measuring retirement income (as opposed to all income), the impact was even more significant, as it increased 27.9 percent under the new questions, or $71 billion.

While having a clearer and more accurate picture of retirees’ income is not doubt valuable to policy makers, EBRI found that by an large, retirees remain largely dependent on Social Security, even when all retirement assets are accounted for.

For 60 percent of retirees in the two lowest income quartiles, Social Security accounts for more than 90 percent of retirement income.

Even those retirees that fell into the second wealthiest quartile get more than 68 percent of their income from social security.

Predictably, those with 401(k) an IRA income were more likely to reside in the upper two income quartiles, proving, yet again, that those Americans with access to workplace savings plans, and those that utilize them, have higher retirement incomes.