social security in the case-deaton era

There is something about the Thanksgiving season that drives me to post things that are a bit random and more or less out of my wheelhouse. Maybe it’s anticipation of the turkey coma to come.

Anyhow, in that spirit, there’s something about the Case-Deaton paper on how middle-aged white people are dying at increasing rates that has been niggling at the back of my mind all month.

The paper, of course, got a lot of attention in the media and blogosphere (including a nice catch by Philip Cohen on how much of the finding is accounted for by changing age composition of 45–54-year-olds). But it’s really the less-educated  whose mortality is increasing, not the whole white population.* And the general finding that inequality in life expectancy between rich and poor is increasing in the U.S. is not particularly new, although the finding of actual declines in life span for some groups is relatively recent.

Obviously, the fact that people in the top income quintile are now expected to get a dozen or so more years of life than those in the bottom — a gap that was a third of that three decades ago — has all sorts of policy implications. But it made me think about Social Security in particular.

Social Security is, on the one hand, a political success because it’s a (near-) universal entitlement program. On the other, there have long been complaints the Social Security is itself regressive, since no Social Security tax is paid on income over $118,500. Of course, Social Security also replaces a larger portion of pre-retirement income for lower-income Americans than it does for higher-income Americans. It’s actually surprisingly difficult to figure out whether, on balance, it’s progressive or not.

What is clear, though, is that if low-income folks are losing years of life while high-income folks are gaining them, the system is losing progressivity (or gaining regressivity). I thought I’d play around with some basic numbers to try to examine this question. But it turns out I don’t need to. The National Academies of Science came out with a big study looking at this only a couple of months ago — a study which, so far as I can tell, got nothing like the media coverage that the much simpler Case-Deaton study received.

So what’s the scoop?

Well, as usual with social policy, a lot depends on the assumptions you make. But making some fairly reasonable assumptions, the NAS report finds that yes, the growing mortality gap is also increasing the gap in Social Security benefits received between low- and high-income groups. Lifetime benefits to the lowest income quintile remain about the same for men born in 1960 as for those born in 1930, and for women they decrease nearly 20%. For the top quintile, though, they increase: about 13% for women, and nearly 30% for men, a huge jump.


FIGURE 4-5 Average lifetime Social Security benefits for males (in thousands of dollars).


FIGURE 4-6 Average lifetime Social Security benefits for females (in thousands of dollars).

So what does this mean? Well, if you see this decrease in progressivity as a problem, it suggests you pay attention to the distributional consequences of various proposed Social Security reforms — which are often not taken into account in discussions of their effects. And it’s not always obvious which reforms will have which effects on progressivity. Raising the early retirement age from 62 to 64 makes the system less progressive, which makes sense. But raising the normal retirement age to 70, though it reduces benefits overall, actually (and unexpectedly) makes things more progressive. And reducing Social Security payouts to those with higher incomes accomplishes this even more directly.

More generally, though, this is a reminder that the growing impact of inequality — an impact that results not only in differential material well-being, but in large gaps in actual years to live — has implications far beyond the obvious ones. The growing gap between rich and poor has the potential to undermine the intent of public policies in a whole variety of ways. We ignore this at our peril.

* Caveat: Just as the population of 45–54-year-olds is not the same in 2013 as 1999, neither is the population of adults with a high school degree or less, the population Case and Deaton identify as having the big mortality increase; this group has become smaller over time and relatively more disadvantaged compared to the population as a whole.

Written by epopp

November 25, 2015 at 1:00 pm

3 Responses

Subscribe to comments with RSS.

  1. Interesting post–I also wonder what type of impact a change in the retirement age would have for inequality across cohorts (especially since the younger cohorts are far more racially diverse than older cohorts). It’s possible we could get greater progressivity by income and at the same time increase racial inequality.

    Relevant for your caveat:

    Hendi (2015) recently showed that the Olshansky et al. Health Affairs study cited in the post ended up getting the wrong end of the stick. Life expectancy declined among white women with less than high school, whereas white men in this group experienced increasing life expectancy. Much of that change is due to the changing composition of education categories (and also to changes in education-specific patterns of smoking: If you look at all races, life expectancy increased across the SES-spectrum for both men and women (though the gap between the least- and most-educated has certainly grown).



    November 25, 2015 at 6:09 pm

  2. How can you change the rules mid-stream? If Social Security benefits are reformed and not paid until age 70, or paid on a scale, less to the more wealthy, there is the problem of motivation early in life and the chance of more free riders – those who work less but still receive the minimum benefit. The problem should be addressed in terms of both employment and healthy lifestyle. Quantitative analyses are fun but does not address the reasons why and ends up with policy changes that cause strategic behavioral changes in the population which searches for the greatest gain in the system. The rules should be left alone and instead the government should address full employment and lifestyle choices. If you want to know why some groups of people have a shorter lifespan, look at the everyday competition.


    Fredrick Welfare

    November 25, 2015 at 6:54 pm

  3. @Arun, thanks for the additional links, and good point about the possibility of becoming more income-progressive but in a racially inequitable way. Another weird thing about evaluating Social Security progressivity is that much of the redistribution is from higher-earning husbands to their lower-earning wives, not from higher- to lower-earning households. I don’t know how much work has been done to disentangle those fairly different types of redistribution.

    @FredrickWelfare, I agree that the incentives created by SS also matter. Congress actually just closed a couple of loopholes that allowed people to game the system a bit:

    But Social Security has been changed a lot of times over the years, and it’s hard to imagine how you could avoid needing to change it over the 40 years or so someone is in the workforce. I wouldn’t say change it right now solely because of growing lifespan inequality, but I would certainly say that next time we decide to change it, this should be taken into account.



    November 26, 2015 at 1:58 am

Comments are closed.

%d bloggers like this: