Older adults in the US may be making nearly one-third more, on average, than commonly thought, according to an important new study by two US Census Bureau researchers. Their report, published last month, finds that the median household income for those 65 and older was about $44,400 in 2012, significantly higher than the $33,800 previously reported by Census.

The implications of the study, by Adam Bee and Joshua Mitchell, are extremely important. It found that, based on these new measures, the poverty rate among seniors was about 6.9 percent, much less than the official 9.1 percent rate. And the importance of Social Security to many older adults may also have been overstated. Using traditional Census measurements, Social Security lifted about 15 million seniors out of poverty. These new data suggest the retirement program kept about 10 million out of poverty.

The study also finds that an often-quoted statistic about the important of Social Security may be overstated. It is often said that one-third of Social Security recipients receive at least 90 percent of their income from the program. But the study found that a big chunk of that is from Social Security Disability Insurance, not from the old age portion of the program. After adjusting for the extra SSDI income, the study concludes that only about 18 percent of Social Security beneficiaries get 90 percent of their income from the retirement program.

What accounts for this significant increase in the income of seniors? The study did not find any meaningful rise in their actual income. They didn’t get better paying jobs or receive retirement windfalls. Rather, the gap was in how they reported their income. This is a bit complicated, so bear with me.

Traditionally, the Census relies on self-reported income. Its benchmark Current Population Survey asks people many questions about their lives, such as health and marital status, living arrangements, school enrollment…and income.  Each year, it does an Annual Social and Economic Supplement (ASEC) survey that it uses to develop income and poverty statistics. Seventy-five thousand people are asked how much they make. And the survey assumes they are answering accurately.

For some time, Census officials and outside experts have been concerned that older adults underreport their income, especially from retirement plans. Now, for the first time, Bee and Mitchell linked the survey results with records from the Social Security Administration and with sample tax return data.

What did they find? When it came to Social Security, wages, and even interest and dividends, there wasn’t much difference between what people told Census and what they reported to the IRS. But there was a big gap when it came to retirement income.

In effect, seniors told Census that they received far less pension and other retirement plan income than either they or the payers reported to the IRS. Two key pieces of evidence:  Form 1040s and the IRS Form 1099-R that shows the amount of pension income or retirement plan withdrawals each year. Bee and Mitchell carefully excluded lump sum distributions and roll-overs from their analysis.

Not surprisingly, the biggest changes–as measured in dollars—were among the high-income households. For instance, the lowest income 10 percent of household underestimated their retirement income by about $1,500 while median income household lowballed by about $10,000 and those in the top 10 percent by about $15,000.

The researchers caution that their study has two important limitations. First, it applies to older adults in 2012 and does not suggest that this trend will continue into the future. Indeed, as fewer Americans receive traditional defined benefit pensions, that unreported income may become less important.

Second, it measures income only and does not look at costs in retirement, debt levels, or other important issues when considering the financial well-being of old adults.

Bee and Mitchell don’t try to explain why older adults underestimate their retirement income. It may be that they just get tired of answering questions and, when asked about retirement income, it is easiest to say “none.” Others have speculated that Americans are less aware of regular retirement distributions because they are direct deposited into bank accounts: You don’t see a check, the income doesn’t register.

There is an argument among economists about whether distributions from retirement plans should be counted as income at all. But without getting into that theological debate, it does seem that seniors have had more cash available for consumption than the government thought.