Often I see proposals for means testing entitlement programs. For instance, Richard Posner writes:
Perhaps some politician will be bold enough to advocate that all entitlements programs, including social security as well as Medicare, be means-tested, as Medicaid is. There is no reason why people who can afford to provide for their retirement should be subsidized by the government, which is to say by the taxpayer. But such a reform does not appear to be politically feasible.
This is a very common sentiment: in an era of impending fiscal crisis, why should the government be paying Social Security benefits to Bill Gates? Economically, however, its logic is questionable.
Means testing is an implicit tax, with all the distortionary effects of ordinary taxes. If $1000 in additional income causes you to lose $100 in government benefits, your effective marginal tax rate is 10% higher than it otherwise would be. It’s strange to see people who would cringe at higher-bracket tax increases suddenly turn around and extol the virtues of means testing Social Security—and it’s hard to attribute this to anything other than a superficial understanding of what “taxes” really are.
But in fairness, the fact that means testing is an implicit tax doesn’t mean that it’s necessarily a bad idea. Two circumstances come to mind.
First, maybe the program in question is highly inefficient. Tyler Cowen, for instance, argues that since Medicare is an in-kind benefit while Social Security is a cash transfer, we should always cut Medicare first. If you believe that Medicare is extremely wasteful, you can make the argument that the benefits from a cutback outweigh the drawbacks of what Greg Mankiw once called “an income tax surchage levied only on old, sick people”. Of course, it’s hard to see how this argument could possibly apply to Social Security, which is as close to a pure transfer program as you can come.
Second, maybe the implicit tax is a relatively efficient one. Under many proposals for Social Security means testing, this is clearly not the case: cutting back benefits on the basis of current ability to pay is a terribly inefficient proposal, wrapping an income tax and a severe capital tax into one. As Scott Sumner once pointed out, this proposal would place the burden of fiscal adjustment entirely on responsible savers like himself, offering continued benefit checks to the former NFL star who blew his savings by his 40th birthday but not to the far lower-income nurse who steadily contributed to her 401(k).
Still, it’s conceivable that we could design a means testing scheme that worked more effectively. For instance, we could means test on the basis of lifetime income. This would mean an effective hike in income taxes—which makes you wonder why we shouldn’t just enact a direct increase instead—but at least there’s a glimmer of an argument as to why it would be efficient. After all, a progressive tax on lifetime income is more distributionally accurate than a progressive tax schedule on year-to-year income: your earnings across several decades are a better indicator of your ability to pay than your earnings in any single year, and there’s no reason to let a millionaire pay at a low rate because he happened to have a bad year.
Moreover, given the current system for calculating benefits, at the margin means testing would be easy to implement. Right now, the Primary Insurance Amount formula for Social Security offers benefits at a rate of 15% for average indexed monthly earnings over $4,517 (and a higher rate for earnings below that). We could easily lower this percentage to 0%, or even a negative rate at sufficiently high levels. Making the PIA formula more progressive is, in fact, one of the recommendations of the much-maligned deficit commission:
The Commission recommends gradually transitioning to a four-bracket formula by breaking the middle bracket in two at the median income level ($38,000 in 2010, $63,000 in 2050), and then gradually changing the replacement rates from 90 percent, 32 percent, and 15 percent to 90 percent, 30 percent, 10 percent, and 5 percent.
But regardless of the specific proposal, there’s one fact we can’t escape: means testing is a marginal tax increase. As such, it ought to be given the same consideration as any other tax increase, and we shouldn’t let a highly distortionary change in incentives slip by just because we’re not officially calling it a “tax”.