November 30, 2015
Andrew Biggs has a piece in Forbes arguing that the standard estimates of retirees income are flawed because they ignore payouts from defined contribution (DC) accounts like 401(k)s and IRAs. Biggs has a point. There is a fundamental asymmetry in the treatment of traditional defined benefit pensions, which send retirees a check every month, and defined contribution pensions from which retirees must make withdrawals. The checks are generally counted as income on our surveys, the withdrawals often are not.
For this reason Biggs is correct to note that measures derived from the Current Population Survey (CPS), which the Social Security Administration uses for its Income of the Aged report, are likely biased downward. The question is how large the bias is. Based on IRS data, Biggs calculates that the correct number for retiree income might be more than 80 percent higher than the income reported by the CPS, an average understatement of almost $6,000 per person. That would be real money.
There are three reasons to think there might be less here than Biggs suggests.
1) Biggs used 2012 as the basis for his calculations, since this was the most recent year for which the IRS provided data on the over 65 population. It turns out that 2012 is a bad year from which to make extrapolations for reasons that every good tax-hating right-winger should know. The tax rate on high-income households was raised in 2013. This means that if you were one of those households, you would probably have wanted to take more from your IRA in 2012 at the lower tax rate.
If we look at the overall taxable withdrawals from IRAs, there was a drop from $230.8 billion in 2012 to $213.6 billion in 2013. Biggs’ extrapolation would have shown an increase in 2013 to roughly $242 billion, an overstatement of more than 13 percent.
2) A withdrawal from a retirement account is not exactly the same thing as income from a DB pension. To see this point, imagine you have $200,000 sitting in an account that you plan to use to support yourself in retirement. It’s possible that you are a careful person and have calculated with reasonable accuracy the amount you can withdraw each year and end up with nothing on the day you die. In this case, your withdrawals will be very similar to the income from a traditional DB pension.
But suppose that you did your math wrong and withdraw the money too quickly, so the $200,000 only lasts half way through your retirement. Then counting these withdrawals the same as pension income would be hugely exaggerating your income. The same problem would occur if you made a withdrawal to deal with a one-time expense, like a high medical bill or a major house repair. Furthermore, some withdrawals may have nothing to do with spending. People may withdraw money to have a different mix of investments or possibly to give money to a child or other relative.
There can be an issue of people withdrawing too little as well, but it’s not clear how to try to weave this into an income concept. It essentially means that people have money to give to their heirs.
3) The third and most important point is that the holdings of IRA assets are highly concentrated among the wealthy. The IRS data show that for the population as a whole, taxpayers with more than $100,000 in income (less than 15 percent of all taxpayers) got almost half of taxable withdrawals in 2013. The analysis of data from the Survey on Income and Program Participation (SIPP) to which Biggs points readers finds that just 19 percent of households over age 65 got any money from withdrawals from defined contribution accounts in 2009. And the distribution of this 19 percent was highly skewed.
Among the bottom quartile, only 8 percent had any withdrawals, for the second quartile the share was 18 percent and for the third quartile 25 percent. The median withdrawals for these groups were $2,200, 2,800, and $3,700, respectively. This means that 96 percent of households in the bottom quartile withdrew less than $2,200 in 2009 from a DC retirement account. For the third quartile 87.5 percent withdrew less than $3,700 from a DC account.
The implication is that Biggs is correct, the standard measures undercount income by their treatment of withdrawals from DC accounts. However for the vast majority of the older population, this is not likely to have much effect on the measurement of their income.
Comments