Income was Too Good to be True

The Census Bureau reported that median income jumped 5.2% last year. As I pointed out last week, that’s a solid increase, but still leaves us short of the record in 1999.

Still, something about the gains struck me as wrong. Even with cities, states, and some businesses bumping up their minimum wage, 5.2% seemed like a lot.

So I went digging. And the more I dug, the more disappointed and frustrated I became. As usual, the source of my angst was the federal government.

Long-time readers know my ongoing annoyance with the Bureau of Labor Statistics (BLS) when it comes to unemployment figures and their birth/death adjustment. Now the Census Bureau is getting in on the act, estimating numbers where respondents don’t give answers.

But the bureaucrats don’t stop there. In addition to filling in blanks, they’re also double counting income from previous years, and the problem is about to get much worse.

In previous surveys, respondents were asked if they held certain types of accounts or received different streams of income, including wages, interest, pensions, government assistance, etc. The details provided were pretty good, but the statisticians suspected they were missing some buckets of cash.

So in 2013 they tested a new set of questions, and compared the answers to the traditional survey. Sure enough, there was more money! By their calculations, incomes were a full 3% higher than had previously been estimated.

The median American family didn’t bring in $51,939 as reported. Instead, that household actually received $53,514. An extra $130 per month could help a lot of families – from those with kids to retirees – make ends meet… if only it was real cash.

To arrive at the higher figure, the Census Bureau parsed questions on items like interest income to figure out if consumers held any interest bearing accounts. Then surveyors would circle back and ask respondents how much they earned on each account. If respondents didn’t know, or refused to answer, the Census Bureau would ask the size of their assets, and then estimate the interest they should have earned.

According to their calculations, by changing the questions the number of Americans earning interest shot up from 86 million to 122 million, and the total interest received in 2013 increased from $187 billion to $387 billion.

The new calculation shows 41% more people earning interest, and 49% more interest earned. Imagine the government’s surprise to find that we’d been holding out on $200 billion in interest income!

To put it mildly, I’m a bit suspect of the government’s new estimate.

But I have a bigger problem with how they’re treating retirement accounts, and what it means for future reports.

Using the old methodology, in 2013 the Census Bureau reported that one million people received income from a retirement account, totaling $18 billion. With the new approach, these numbers jump to 5.2 million people and $60 billion. The higher numbers make more sense to me, given that our over-70 population keeps growing, but all the funds withdrawn aren’t income.

The government asks if funds withdrawn are reinvested elsewhere, taking care of that part of the equation. My issue has to do with the original money.

The survey counts as earnings all wages, salaries, and money from almost every source that isn’t borrowed, or insurance proceeds before paying taxes, contributing to pensions, paying union dues, etc. So the money workers earn before contributing to a retirement plan is counted as income in the year it was received. And then it’s counted again in the year it’s withdrawn from the retirement account.

Hmmm.

With so many boomers funding retirement accounts over the years and now retiring in large numbers, this problem is going to get worse. We’ll have the Census Bureau telling us that income is on the rise, when it’s really just savers withdrawing the funds they contributed.

Interestingly, this is not how the bureau treats other accounts. Funds withdrawn from savings accounts aren’t counted as income, just the interest is, which makes sense. The original deposits were earned at some previous point.

The obvious solution for the double counting is to calculate how much of the funds withdrawn from retirement accounts represent earnings and how much represents investment and interest income.

Obviously that’s a big, hairy question, but hey, I’m not the one double counting, they are.

Then again, I’d imagine the government has no interest in a “solution.” They’ve already found one. Their redesigned survey shows income moving upward and onward, and a positive impression is all that matters, right?