Anne Alstott spoke at NYU Law School today regarding her new book, Social Security, A New Deal for Old Age: Toward a Progressive Retirement.
I haven't read it yet, but plan to. It contains a detailed plan for modifying Social Security that, for many people, will be its chief takeaway - including, for example, a proposed floor on benefits and gradual removal of the "free" secondary earner benefit, grounded in discussion of economic and demographic changes over the last few decades. These are certainly changes that I would view sympathetically.
But I'm a bit of a quirky reader - I tend to be more interested in useful conceptual tools that help me to advance my understanding than I am in reform specifics. From that standpoint, here was my favorite takeaway from the talk. The book proposes using information that the Social Security system has regarding one's career earnings (or at least covered earnings) through age 62 to affect the relationship between annual benefits under early, normal, and late retirement. Or more specifically, rather than just having the three options of early, normal, and late like current Social Security, it proposes a gradient of annual benefits, for any given career earnings level, depending on which retirement year one chooses as between ages 62 and 76. But onto the conceptual part.
Here are two things that are generally known among people who discuss Social Security policy. First, while high-earners often have the types of jobs in which they can easily keep going long past age 65 or 70, this is not so true for low-earners, who often really are physically constrained in a new way, relative to their jobs' requirements, by the time they reach the early 60s. Second, life expectancy is now significantly greater for high-earners than low-earners.
Both points show (as has been previously discussed by various analysts) that it is not so obvious that we should generally raise the Social Security retirement age, even if average life expectancy is increasing. But why stop the analysis there.
The first point also shows that low-earners with inadequate retirement savings may need to choose early Social Security retirement even if it is actuarially bad for them. The second point shows that, if one wants the tradeoff between early, normal, and late retirement to be actuarially neutral as to a given individual, one needs to distinguish between low-earners and high-earners. They will have different neutrality points, so far as the tradeoff between annual benefit levels and the expected lifetime value of the retirement benefits is concerned.
Alstott proposes to have the decline in annual benefits, as one chooses between retirement at age 62 or 76 or somewhere in between, depend on career earnings. Low-earners get a more favorable tradeoff, so that it's easier for them to retire early in both an absolute and relative sense.
Conceptually, I think of this as using Social Security's information about career earnings in a new way. The fact that a given individual had low, rather than high, career earnings is evidence, statistically speaking, of both (a) a drop in current-period earnings "ability" when one reaches the early 60s, and (b) a low individual life expectancy, which is relevant to the actuarial neutrality point from retiring early versus late. This is information that the system can use creatively in benefit design.
With rising income inequality producing (and/or being produced by) rising heterogeneity as between members of the workforce at the high end as compared to the low end, it's increasingly important to use such differentiating information.