Wednesday, August 5, 2009

New paper: “Should Social Security rely solely on the payroll tax?”

The Center for Retirement Research at Boston College has released a new paper by Alicia Munnell, "Should Social Security rely solely on the payroll tax?" Here's the summary, followed by some comments:

It's no secret that Social Security is facing a long-term financing shortfall.  This problem can be solved only by putting more money into the system and/or by cutting benefits.  There is no silver bullet.  So the following discussion is not to suggest that there is an easy way out, but rather to explore whether the entire financing of the Social Security system should rest on the payroll tax.  The payroll tax may be a perfectly reasonable way for current workers to pay for their benefits.  But is it the right tax to finance the costs left over from paying benefits far in excess of contributions to early generations?  

This brief explores the question of the appropriate tax, or combination of taxes, to finance Social Security.  Since the need for more revenues gives the question increased currency, the first section briefly describes Social Security's financial outlook.  The second section then describes the payroll tax.  The third section explores whether the whole cost of the Social Security system – the contributions necessary to generate current benefits and the contributions required to make up for giving early participants benefits far in excess of their contributions – should be financed in the same way.  The fourth section concludes that perhaps a portion of Social Security financing could be transferred from the payroll tax to the income tax.  It would mean higher income taxes, but the burden of the "legacy debt" would be borne more broadly.  

My thoughts: Alicia rightly differentiates between the future costs of paying benefits and the "legacy costs" leftover from the fact that Social Security paid most early participants significantly more in benefits than those participants paid in taxes. (Here's a chart, generated from SSA research by Dean Leimer, showing rates of return by birth cohort. When the rate of return is above the trust fund bond rate, then from Social Security's pleasant of view that person or birth cohort is a money-loser.)

I also agree that there are merits to financing these legacy costs separately from the rest of Social Security. This would, in effect, create two forms of Social Security taxes: the "legacy tax," upon which no benefits would be paid, and the "Social Security contribution," which would be repaid in full through future benefits, including interest.

Where I'm not sure I agree with Alicia is on her argument that these legacy costs should be financed according to the "ability to pay," which she interprets as meaning through the income tax system. This would involve increasing income taxes by around 15 percent, depending on the period over which we wished to repay the legacy debt.

But from a point of equity, the legacy costs of overgenerous benefits were generated according the Social Security tax and benefit schedule and it's not clear to me that simply levying a payroll tax to finance the legacy costs is particularly unjust. Low and medium earners did and do benefit from Social Security's progressivity and there's no reason they shouldn't play a part in financing the costs of putting the system on a sustainable track. A payroll tax is based on the ability to pay – the higher your earnings the more you pay. Income taxes would primarily hit people earning over the Social Security tax/benefit cap, which seems a somewhat tenuous relationship to how the legacy costs were generated.

More broadly, I'm not sure the financing of the legacy debt should be decided based on fairness criteria rather than questions of efficiency. At the margin, the most efficient means of raising new revenue may well be a consumption tax such as a sales tax, VAT, carbon tax, and so on.

In any case, though, this paper presents interesting and useful new ideas and is worth checking out.

1 comment:

JG said...

This seems to get very much back to Milton Friedman's argument of a decade ago -- not noted by Munnell.

Friedman proposed considering what would happen if we simply ended the current SS system on an arbitrary date ... then paid all benefits earned under it until then using general revenue ... and then started a fully funded system for workers from that date on (or didn't start a new system, it doesn't matter for the analysis).

His first point was this shows the claim that a "transition cost" must result from converting from the status quo to an alternatve private system is bogus, a myth. The purported "transition cost" exists of financing the unfunded legacy obligations of the status quo (as clearly seen with the imposition of the old-program cut-off date). Since that cost exists in the status quo, and it's financing is the cause of all the current program's problems -- underfunding, declining returns on contributions etc. -- it can hardly be fairly described as a "transition cost" resulting from moving to a different system.

Beyond that he argued it makes more sense to finance the legacy cost through general revenue, because the cost of the "excess" benefits to past retirees was an expense the nation as a whole chose to incur, for good or ill, comparable to other national expenditures such as for World War II, the Apollo moon program, etc. -- which the nation did not finance through payroll taxes imposed on a limited wage base.

E.g., funding the legacy expense from general revenue would free the worker retirement progam to become an attractive pension scheme again, with higher returns on lower contibutions -- a progressive improvement. It would also fund the legacy cost on a much larger tax base, allowing a lower tax rate -- an efficiency improvement, and a progessive improvement too since income tax hits wealthier taxpayers.

Beyond that he said the political fixation on payroll tax financing that exists nonetheless, in spite of all the above, shows the great mesmerizing, politically paralyzing power of the system FDR set up, even when it doesn't make much sense. The great power of path dependence in politics, some might say.

Or as Friedman himself put it:

"The link between the payroll tax and benefit payments is part of a confidence game...."

Low and medium earners did and do benefit from Social Security's progressivity and there's no reason they shouldn't play a part in financing the costs of putting the system on a sustainable track.

Today's low- and medium-earners do not gain from getting any of the "excess" above-market returns, so what do they have to pay back?

As to giving the early cohorts of retirees those excess returns, Munnell says "Virtually all observers agree that the decision ... was a wise one."

Maybe it was -- but today's low and medium wage workers weren't a party to that decision and don't benefit from it, so why should they bear such a disporportionate share of the cost of it? I'm with Friedman in thinking this cost should land on the entire national fisc, not just a limited payroll wage base.

At the margin, the most efficient means of raising new revenue may well be a consumption tax such as a sales tax, VAT, carbon tax, and so on.

An interesting point may be that if one concludes that efficiency-fairness-progressiveness dictate that general revenue (income tax, VAT, carbon tax, whatever) is a superior option to payroll tax as a source for additional revenue for Social Security, then one must ask why it isn't also the superior option for collecting its current level of revenue.

If it is the superior option for the next marginal dollar of revenue, why not for the preceeding dollar? If it's a better option than collecting a 14th percentage point of paryoll tax revenue, why is it not better than collecting the 11th?

Is the payroll tax that mesmerizing, political path dependence that strong?