Here’s a letter Yglesias and I both got in response to my Social Security column yesterday and Matt’s comments at TalkingPointsMemo:

A reductio ad absurdum argument that I believe strongly counters Mr. Sanchez’s argument that Social Security not be considered an insurance program, mentioned critically by you in TPM:

“Neither is the notion of health insurance as ‘insurance’ terribly coherent, if it ever was. Even when health insurance was first instituted, over half of Americans who reached the age of 21 would need health care. As of 1990 the percentages were over 99.9 percent for men and 99.9 percent for women. Needing health care is not a ‘risk’ to ‘insure’ against; it’s a normal part of life to plan for.”

A slightly less shoddy version of the quickie reply I shot back:

Look, as a semantic matter, you can call any transfer program “insurance” if you want. (Food stamps are hunger insurance, TANF is single-motherhood insurance). There are two distinct relevant questions. The first is whether “insurance” (as opposed to a simple social benefits program) makes sense as a metaphor for the SS system as currently structured. The second is whether an insurance model makes sense for thinking about retirement generally. And the answer to both is, I think, “no.”

On the first: Social security’s got the slightly insurance-like feature that payouts are tied to payroll taxes. But it’s well established as a matter of law that benefit outlays are at the discretion of Congress. They’ve been changed before and could be again, and the Supreme Court’s established that there’s no legal obligation to pay anyone any particular benefit. So on that score it sure looks like any other social program. Neither is there any attempt to hedge risk in the way an actual insurer would—men and women pay the same payroll tax, f’rinstance, despite different life expectancies. So the current structure of SS has some superficially insurance-like features, but bears a much greater resemblance to an ordinary tax and transfer program.

The second question is the one Matt addresses, and he makes it sound plausible because there are obviously aspects of old-age that do make sense on the insurance model. Disability insurance, for instance—a hedge against the risk of being rendered incapable of working—is appealing on its own terms, regardless of one’s age. And people routinely hedge against outliving their nest-eggs after retirement by purchasing private annuities. The question then is why, apart from the redistributive aspect, there’s any reason to federalize the risk-hedging component of retirement planning, rather than letting individuals do that, factoring in health, the kind of work they do, the kind of work they’d be capable of doing if they stopped doing that job, how much they want to work for how long, and so forth. Since that’s highly sensitive to personal situations, a savings model with some individually-tailored insurance component makes sense. The only rationale for government involvement is the redistributive component: Providing a baseline for those who aren’t able to save enough to hedge appropriately. And, again, you certainly wouldn’t start the payouts at 65. You might insure against living unusually long, but you wouldn’t shift from savings to insurance that early on any more than you’d insure against the need to pay for food and rent next month. (It is a risk in some sense: You could be hit by a car tomorrow, eliminating the need to pay for either.)

Not to get too far afield, but the healthcare example actually points out a systematic irrationality in our healthcare system—paying for routine checkups through insurance rather than out of pocket—that’s largely an artifact of tax policy, though part of the logic is that if the insurer is going to be liable for big risky expenses—which are the ones it makes sense to actually “insure against” on their own terms—they want the incentive structure to be such that you’re going to see a doctor routinely and lower their liability over the long term by catching problems early. But to generalize from that and assume that health is a “counterexample” to the notion that you don’t, as a rule, insure against things that are near-certainties won’t work.

So, fortune cookie–sized summary: The social purposes of government involvement in retirement, which is to say, the redistributive aspects of Social Security, are served by a welfare program. The risk-hedging purposes would be better accomplished by individuals deploying their local knowledge, and very few sane individuals would actually structure their risk this way, with an “insurance” benefit kicking in at an age past which the vast majority of folks will predictably live (and therefore either keep working or plan to have savings). What Social Security does is provide for everyone an insurance contract that many people would be perfectly capable of tailoring and purchasing on their own in order to make less obvious the subsidy to those who couldn’t save and purchase that service on their own. But the subsidy part of it is the only reason to get government in the game.

Now, Matt’s comments seem to recognize this well enough, since he describes Bush’s proposal as an attempt to disaggregate the functions, and cites largely political reasons for keeping them linked together. But—as I hope I’ve at least sketched out—there are reasons to keep ’em separated, and we surely would if liberals weren’t worried the redistributive component would get slashed if it were out there on its own.

I don’t want to suggest there’s no genuine tension here: If we turned Social Security into a means-tested old age poverty assistance program, he’d probably want a larger benefit with fewer conditions than I’d like to see. But I wonder whether the fear isn’t overstated. Consider: What are the most potent arguments Democrats use now in defense of Social Security? They’re mostly the arguments that’d apply to a welfare system: We don’t want to see people who worked low-income jobs their whole lives mired in poverty when they’ve grown too old to work. Well, that’ll still hold under a disaggregated system. Would I like them to rely more on family or community organizations or private charity and less on government to the extent possible? Sure. But the public at large isn’t full of wacko libertarians like me.