Cross-posted from AEI and Authored by Andrew G. Biggs
A recent Bankrate.com analysis claimed to show “America’s widespread lack of retirement preparedness at a time when many people have a hard time planning even week to week, let alone for an event that is decades away.” Bankrate found that retirees had average incomes equal to just 60% of working-age households, 10 percentage points short of the 70% “replacement rate” that most financial advisors recommend. In only three states – Hawaii, Alaska and South Carolina – did Bankrate find that average retirement incomes topped 70% of pre-retirement incomes.
So are things really that bad for today’s retirees? Not at all. In fact, if we recreate Bankrate’s analysis while correcting for several important errors, today’s retirees don’t merely meet the 70% “replacement rate” target – they actually have average incomes that match those of households who are in their peak earning years. Even if Americans had saved perfectly for retirement, one would expect retirees to have much lower incomes than working-age households. So income parity is a surprising and reassuring result.
To start, Bankrate measured retirees’ incomes using the Census Bureau’s American Community Survey. While a fine dataset for many uses, the ACS misses most of the income retirees collect from IRAs and 401(k)s, as well as from lump sums paid by traditional pensions. Like Census’s Current Population Survey, the ACS counts money as “income” only if it’s paid in regular amounts, like a weekly or monthly check. But most people draw down their IRAs and 401(k)s on an as-needed basis, so – believe or not – that money doesn’t count as “income” in Bankrate’s calculations. As retirees more and more rely on IRAs and 401(k)s, the ACS will increasingly understate retirees’ incomes.
Second, retiree households are smaller than working-age households, which are generally married couples with children. Based on the Survey of Consumer Finances, households ages 45 to 64 have an average of 2.3 people, versus just 1.7 for households aged 65 and over. In comparing incomes for households of different ages, we need to remember that smaller retiree households mean higher per capita income.
And third, what matters to retirees isn’t matching the income of working-age households, but maintaining the standard of living that retirees themselves had when they were working. This is what financial planning tells us, as well as the more sophisticated “life cycle model” used by economists. If there were zero growth of household incomes over time, this difference wouldn’t matter. But when incomes rise over time – and according to the SSA, the real average wage has risen by about 27% over the last 21 years – then comparing retirees’ incomes to those of working-age households will be misleading. If real wage growth is positive, then retirees should have lower incomes than working-age households, even if those retirees – for some reason – wanted a retirement income that perfectly matched their pre-retirement earnings.
So let’s take another stab at this. If we want to accurately measure retirees’ incomes we need to accurately capture IRA and 401(k) withdrawals. That points toward IRS Statistics of Income data, since those withdrawals are taxable and thus are more fully reported. For households aged 65 and over in 2013, IRS data show an average total income of $70,085. Divide that by 1.7 individuals per household and you get per capita income of $41,227. Those same IRS data show an average income of $88,670 for households aged 45 to 64. Divide that by an average household size of 2.3 and you get per capita income of $38,552. While the IRS data have some shortcomings, which I’ll discuss below, these figures don’t look anything like the 60% “replacement rate” the Bankrate analysis produced. Indeed, today’s retirees have incomes that are very similar to Americans who are in the prime of their working lives. And that’s before accounting for the increase in salaries from the time today’s retirees were in their prime working years.
But about the IRS data’s shortcomings – there are several. For instance, households with very low incomes don’t have to file returns, so tax returns aren’t a complete picture of either the working-age or the retired populations. Likewise, the IRS SOI data give us only averages (means), when we’d probably prefer to look at medians. The ratio of median retiree income to median working-age income could be higher or lower than the ratio of means, depending upon whether incomes are more unequally distributed among retirees than among prime-age workers. Third, the IRS has not released income data by age for years prior to 2006, so we can’t easily go back to see the incomes today’s retirees had back when they were working. Fourth, while nearly all married couples file joint tax returns, for the roughly 5% who file separately the IRS data will treat them as a separate household. So while the IRS data has a crucial advantage over datasets that effectively ignore IRA and 401(k) income, that advantage does come with some downsides.
But still, you get the point: Fully counting the incomes received by today’s retirees – especially, withdrawals from retirement plans – shows that, on average, retirees are doing very well. No one argues that retirees need per capita incomes as high as households 20 years younger who are at the peaks of their working careers. But according to the IRS data, that’s what they have.
Does that mean all retirees are doing fine? Of course not, nor are all working-age households doing well. But if we’re looking at averages – and averages do tell us something meaningful – then today’s retirees are doing pretty darn well.
Cross-posted from AEI and Authored by Andrew G. Biggs
Bold TV is a digital news network committed to bipartisan dialogue & innovation for people, business & communities.