Interesting paper today, and one of broader policy interest, at today's Tax Policy Colloquium, which I am running along with Alan Auerbach at NYU Law School this semester.
BTW, the weekly schedule and papers are available on-line here.
Furman addresses what one might call the second-order fiscal problem in Social Security. The first-order problem is the program's existing fiscal gap, recently estimated at about $11 trillion. (And of course the broader U.S. fiscal gap, without which this would be less of a concern.) The second-order problem is that, even if we fixed the first-order problem based on our best estimates as of today, a problem of under- or over-funding would reemerge once there was new information about demographics, wage growth, etc. We can probably be confident that the political system will be unable to handle these changes in a timely fashion, for a range of reasons ranging from voters' loss aversion (relative to current law) to interest group politics to chicken games between Republicans and Democrats re. how to adjust the system.
Furman therefore proposes that changes be evaluated in terms of "robust solvency," i.e., how would their restoration of sustaiability survive changes to economic and demographic forecasts. Built-in features of the system reduce the risk exposure to changes, e.g., in GDP growth, but changes in the worker to dependent ratio, e.g., due to changes in fertility or mortality, are a different matter.
An example of a change that would not meet the robust solvency standard is changing from wage indexing to price indexing of benefits, for affluent seniors or all seniors. The problem with this change, from a robust solvency standpoint, is that, if economic growth is lower (increasing the Social Security funding program), the benefit cut it imposes is smaller (since wage indexing is then less in excess of price indexing), while, if economic growth is higher and benefits cuts are thus less needed, the benefit cut is higher. He therefore considers a shift to price indexing markedly inferior to an identical benefit cut, by an expected-value standard, that lacks this perverse and backwards feature.
As an implementation of robust solvency, Furman proposes dependency indexing, or having taxes and/or benefits adjust automatically to unexpected changes in the dependency ratio. This would make the system self-correcting without requiring the political system to show unexpected maturity in responding in a timely fashion. This idea could be combined with any Social Security reform that created sustainability under current median projections.
I consider this a very creative and worthwhile idea that merits serious consideration at a minimum and incorporation into various reform proposals on the high end.