The GAO report, titled “Retirement Security: Shorter Life Expectancy Reduces Projected Lifetime Benefits for Lower Earners,” focuses on how the fact that high earners live longer than low earners partially undoes Social Security’s progressivity. The GAO was responding to a request from Sen. Bernie Sanders in which Sanders, in GAO’s words, “asked us to examine disparities in life expectancy and the implications for our nation’s policies with respect to retirement security.” There’s a lot of detail in the GAO’s report, some of it extraneous and some of which is misleading. But I don’t think GAO really answers the questions Sen. Sanders was asking.
GAO finds, for instance, that differential mortality between rich and poor means that a low-income male collecting Social Security benefits at age 65 will receive 10% lower lifetime benefits than were he to have the population-wide average life expectancy. A high-income male, by contrast, receives 12% higher lifetime Social Security due to his above-average life expectancy. In other words, the lifetime collection of Social Security benefits is less progressive than benefit payments in the first year of retirement, because the rich live and collect benefits longer. This has been known since at least the 1970s, but the fact that the longevity gap between rich and poor has been growing makes it a relevant topic for current research.
That’s an interesting and relevant point regarding Social Security’s lifetime progressivity. But what does it have to do with retirement security? Retirement security is generally measured as a retiree’s ability to maintain his pre-retirement standard of living, the shorthand for which is the replacement rate, meaning retirement income as a percentage of pre-retirement earnings. Differential mortality doesn’t really change that, at least from Social Security’s perspective. The replacement rates that low, middle and high-earning households receive from Social Security don’t change as a result of the fact that high-earning households live longer than low-earning ones do. Yes, the program becomes a bit less progressive on a lifetime basis. But note, it’s a bit less progressive, not regressive. Even with differential mortality, low earners receive more in benefits than they pay in taxes while for high earners it’s the opposite.
But differential mortality does tell us something interesting about retirement security that the GAO doesn’t spend much time on. Most studies on the adequacy of households’ retirement saving will express the balances of savings accounts like IRAs and 401(k)s as an “annuity” at retirement age. An annuity is a financial product that allows you to convert a lump sum of money at retirement age into a monthly retirement income that last for life. The annuity formula used in retirement income adequacy studies is usually based on the average life expectancy of the retiree population. Converting lump sums to annuities is unrealistic – very few people actually purchase annuities – but it’s a convenient method for analyzing retirement incomes.
But it’s problematic when few people actually purchase annuities and when the rich live much longer than the poor. Let’s say that a rich man and a poor man both retire with 401(k)s. In practice, both will gradually draw down their account balances. If, as the GAO notes, the rich man lives 12% longer than average, that means his annual withdrawals must be 12% lower or he’ll run out of money late in life. Likewise, if the poor man’s life expectancy is 10% shorter than the average, he can increase his 401(k) withdrawals by 10% and still not run out of money.
By itself this doesn’t overturn our thinking on retirement saving. Low-earners tend not to have a lot of assets, so adjusting up their 401(k) withdrawals by 10% isn’t going to make them rich. High-earners do rely heavily on 401(k)s and other retirement accounts but, again, they’re high earners. But the simple fact that some retirees’ assets won’t have to last nearly as long as we thought, while others will have to last longer, strikes me as worth mentioning if writing a study on differential mortality and retirement security.
The GAO makes some policy news regarding Social Security reform. They state,
“Our analysis indicates that one frequently suggested change to address Social Security’s financial challenges, raising the retirement age, would further reduce projected lifetime benefits for lower-income groups proportionally more than for higher-income groups.”
I’m going to say flat out that this claim is wrong and GAO should know it’s wrong.
Some people get very confused about how raising the Social Security retirement age affects progressivity, because they don’t realize that raising the Social Security retirement age is nothing other than an across-the-board benefit cut. If the retirement age goes up by a year, then everyone – rich and poor, long-lived and short-lived – receives about 7% less in retirement benefits. They receive 7% less in each year they retire and they receive 7% less over their lifetimes. GAO’s analysts surely understand this.
So how do they get to the conclusion that raising the Social Security retirement age would disproportionately hurt low-earning, shorter-lived retirees? By packaging an increase in Social Security’s “normal retirement age” – currently age 66, which is the retirement age policy that’s actually “frequently suggested,” in GAO’s terms – with another policy that’s much less frequently suggested and which doesn’t contribute to fixing Social Security’s solvency problem: raising theearly retirement age of 62, which is the age at which people can first claim retirement benefits. Unlike raising the normal retirement age, raising the early retirement age is regressive because those who die between 62 and the new early retirement age, say 64 or 65, would not receive benefits. Because of differential mortality, those individuals would be disproportionately poor.
The GAO is handing a talking point to people who oppose raising the Social Security retirement age, but it’s a misleading one. GAO’s results depend on adding a second provision, raising the early retirement age, to the reform package. There are plenty of Social Security reform proposals that raise the normal retirement age but don’t touch the early retirement age. The GAO is unfairly maligning those plans as harming the poor.
I personally favor raising the early retirement age as a way to encourage longer work lives, coupled with other provisions that would make Social Security as a whole substantially more progressive. The GAO acknowledges – in the footnotes – that this is possible. But there’s no reason they shouldn’t have noted, prominently, that the increase in thenormal retirement age is what’s doing the legwork in terms of fixing Social Security’s solvency while it’s the increase in the early retirement age that causes the regressive results.
Finally, for no reason I can figure out, GAO includes an appendix which is nothing other than a chart showing that the share of total earnings subject to taxation had declined since the 1980s. (Compared to Social Security’s full history since 1935, however, the taxable earnings share today is right about average.) The share of earnings subject to taxation has nothing to do with differential mortality, but is a favorite talking point of progressives who want to fix Social Security by lifting or eliminating the so-called “tax max.” That’s fine, but it doesn’t belong in this kind of report.
One of the upsides (for me, at least) of working at a think tank is that people who read my work for the hard analysis get my slanted opinions as a free bonus. But that’s not how it should work for a government agency producing work for the public.