Tax policy colloquium on Lily Batchelder's Savings Incentives with Insurance Objectives: A Bankrupt Approach?

Last Thursday we discussed Lily Batchelder's new draft paper (still a work in progress), which argues that savings incentives in the tax code are poorly designed to advance insurance objectives for poorer individuals. She notes low responsiveness to the rules in present law, and their generally providing larger incentives to higher-income individuals who arguably are saving enough already. Thus, she proposes either (a) changing defaults so people automaticallty save for retirement through their employment, but with opt-out and no tax benefit for saving, or (b) a refundable saver's credit that's phased out based on income. Absent opt-out, the savings accounts would automatically convert shortly before retirement to fixed real life annuities.

Alan Auerbach and I were more inclined than Lily to think of saving enough and having enough insurance of various kinds as very different things. Suppose you know with certainty (a) your entire future earnings profile, (b) your exact life expectancy, and (c) anything you need to know about consumer prices, your consumption preferences, and rates of return on saving until the day you die. In this scenario there's no uncertainty about any of these inputs, hence no need whatsoever for insurance so far as any of them are concerned. Yet it would still be vital to save enough to smooth your lifetime consumption path optimally, and people who were myopic or had self-control problems would fall short of optimizing. Hence, we might want to require or induce more saving in order to increase their welfare (and also to save us from having to "rescue" them later on if they depart too far from the optimal path).

Likewise, consider in these terms Social Security. In large part it is simply a device to force a minimum level of retirement saving given one's lifetime income (net of taxes and transfers, including its own). Indeed, though it is always called social insurance for purely formal reasons (e.g., its purporting to have dedicated financing, which in fact aren't much like insurance premiums), its sole insurance feature (redistribution in the program aside) is the fixed real life annuity.

Suppose that we knew everyone's exact lifespan. We'd still need Social Security, but we'd be able to substitute term annuities, equal in length for each retiree to her remaining lifespan, for the life annuities. Now it wouldn't be insurance at all, yet it would still be pretty similar to what we now have. So insurance is not really a huge part of Social Security, despite the semantic convention that supposes it to be a core case.

If we limited the "social insurance" label to programs that actually are economically insurance, and furnished by the government due either to adverse selection problems or people's under-appreciating the value of being insured, the preeminent case would be the income tax plus welfare system, which provide insurance against income risk. But these of course are the programs that no one calls insurance and that indeed were the very ones the inventors of the "social insurance" label were trying to distinguish from their favored programs. (See my Social Security book for a fuller discussion.)

Anyway, the main payoff of all this to Alan and myself was disagreeing about the extent to which savings incentives and insurance objectives should actually be considered a natural match.

The other main quibble we had pertained to the proposal for default saving with free opt-out. Lily wants the opt-out, rather than mandatory saving beyond that in Social Security, because otherwise we might be requiring someone to over-save relative to what was truly optimal for them. Our objection was that we don't really have good reason to tilt the default towards more saving unless one thinks people are otherwise likely to be saving too little. But if we think they mostly are, why allow the opt-out?

The reason for allowing it would be clear if we expected the "right" people to opt out while the ones we wanted to save more stayed at the default level. But what if it was the other way around? It's hard to know why the opt-out would be exercised primarily by those who actually would be saving too much otherwise, rather than by those who want to under-save relative to what's optimal.