Tuesday, March 22, 2005

Well, At Least He's Engaging

Alex Tabarrok at Marginal Revolution tries to give Senator Lieberman an assist:
Brad DeLong and Paul Krugman are taking Joe Lieberman and others to task for asserting that the cost of fixing the social security problem increases at $600 billion a year. I agree that Lieberman is confusing an increase in the nominal present value of the debt with an increase in the cost of fixing social security but in correcting Lieberman both DeLong and Krugman meander towards the opposite error - that the costs of fixing social security is not increasing.

But almost inevitably a fix to social security will involve tax increases and the longer we wait the larger the costs of those increases will be. The technical explanation is that deadweight loss increases more than proportionately with an increase in taxes. The common sense explanation is that you don't want to take all your hits at once - instead, if you must take a hit, it's best to spread it out over time. Thus, the sooner we deal with the problem the lower the total costs will be. Lieberman's message is
correct, even if the details are wrong.

Yes, the $600 billion number is not the right number. For the right number and the right argument, I conjure my first post on Social Security from last October:
At present, the Social Security actuaries project an unfunded obligation of $10.4 trillion in the Old-Age, Survivors, and Disability Insurance (OASDI) program. This number comes directly from the 2004 Trustees Report released in March. (See Section IV.B.5 and Tables IV.B.7 and IV.B.8 in particular.) This is the present value of the projected payments less the present value of projected revenues for the system over the infinite horizon. It is the most comprehensive way to measure the hole in the system's finances.

Note that this is the unfunded part of the obligations--it is over and above all of the payroll taxes (12.4 percent of taxable payroll) and income taxes on benefits that go to support the program under current law. If this gap were to be closed through payroll taxes, it would require them to be raised by 3.5 percentage points, immediately and permanently, with the additional surpluses over the next few decades saved (in Treasury bonds) to finance annual deficits that are projected to grow to about 6 percentage points of payroll over the next 75 years.

This $10.4 trillion unfunded obligation is sometimes referred to as implicit debt, to distinguish it from the federal government's explicit debt issued in the form of Treasury bills, notes, and bonds held by the public. At present, implicit debt from Social Security and Medicare is several times larger than the government's explicit debt. Is having so much implicit debt a problem? I think so, and the reason is that, just like explicit debt, we accrue interest on implicit debt.


So if we have an implicit debt of $10.4 trillion, and the real interest rate is 3 percent, then next year, the implicit debt will grow by 0.03*10.4 trillion = $312 billion, up to $10.7 trillion, if the assumptions underlying the projection stay the same. Why does this matter? Primarily, it matters because both the President and Senator Kerry have repeatedly stated (see the two speeches in Pennsylvania linked above) that they will not cut benefits for those at or near retirement age. (The Senator's statement may be even more encompassing, including benefits at any time in the future. I cannot tell for sure from his public statements.) This, in turn, means that each year that elapses without reform causes the burden of financing the unfunded obligations to be shifted away from one more birth cohort that crosses the threshold of being "at or near retirement." The more we wait, the larger the burden on future
generations, and the higher that 3.5 percentage point surtax would have to climb.
The $10.4 trillion is about 90 percent of current GDP. In a later post, I made a rough calculation that if we waited until 2042 (the projected date of trust fund exhaustion), the implicit debt would grow (at the 3 percent real interest rate), to about $32 trillion, which would be about 141 percent of that year's (much larger GDP). So even if taxable payroll didn't fall as a share of GDP, the surtax applied in perpetuity would have to increase by a factor of 141/90, or from 3.5 to 5.5 percentage points.

The issue that Alex is pointing out is tax smoothing: for efficiency reasons, it is better to have a surtax rate that is steady at 3.5 percent rather than one that is 0 for 38 years and then jumps to 5.5 percent. The issue is, for me, less about tax smoothing and more about the intergenerational fairness of consigning future generations to pay higher payroll tax rates. We shouldn't be doing that--in Social Security, the General Fund, or Medicare.

Other blogs commenting on this post


PGL said...

There is a serious problem even with your analysis when one thinks about how Bush has restricted what he calls reform. This relates to that odd Bruce Willis tale Paul cooked up. Bush has ruled out any new payroll contributions so the PV of inflows will not go up regardless of when we pass and implement his proposals. His proposals rule out anyone retiring in the next 10 years losing benefits. So if we pass now or we pass something in 2014 that applies to all born after 1950, there is NO difference AT ALL. This is a simple point so I would hope smart and honest conservatives would acknowledge it.

franco said...

In other words, the cost of delaying one year is simply the foregone revenue from taxing people for that year. If a 3.5% tax in perpetuity would put the system in balance, waiting one year would require finding an additional $150B or so, not $600B.

Of course, the best policy change might be to reduce benefits -- perhaps by raising the retirement age with life expectancy. Such a change could conceivably cost even less to delay if it disproportianately hits later generations. Given their expected greater longevity, that might well be very fair.

Finally, a question: how much of the famous $10.4T unfunded obligation is already accrued? That number includes inflows and outflows associated with people who have not yet entered the workforce. How much of the UO comes from them vs people who have already paid some payroll taxes and accrued a promise of benefits?

It seems to me that the right policy ought to depend on that number. Consider the 2 extreme possibilities:

1. Much of the UO is not yet accrued. Basically this means that future cohorts are getting an actuarially good deal, better than fair. Any adjustment should focus on avoiding giving whole cohorts a good deal at the expense of previous generations. If current cohorts are getting a good deal, we should act fast to stop that. A permanent tax hike to pay for this would potentially be an intergenerational transfer from current generations to future generations -- probably something most would agree is undesirable.

2. All of the UO is accrued and then some. Future cohorts are getting an actuarially unfair deal, but the program is benefitting from their participation. In this case, we could break SS into 2 pieces, SS-future ans SS-past. Say the former covers everyone that has never paid anything other than 12.4% or has not yet entered the workforce. All of the UO would be associated with SS-past and there would be no problems (other than excess cash) with SS-future.

This manipulation is just to highlight the possibility (in scenario 2) that the program going forward would be perfectly healthy except for the fact that it is currently assigned the task of paying off the implicit debt we've already accumulated.

Here, it would seem to make sense to have those that contributed to the problem help fix it to the extent possible, and then spread any remaining burden thinly over many/all generations (or simply cover it out of general revenues which amounts to about the same thing).

I'm not trying to express personal views in my discussion of these 2 scenarios, just trying to motivate why my question might be relevant to evaluating policy choices.

PGL said...

franco - very well said. The only way waiting increases the alleged shortfall is that the 55 year old plus crowd does not sacrifice a thing. This 49-year old is willing to sacrifice a few future benefits - and at least I know none of my benefits will be paid to me until circa 2020. So the point of Joe Lieberman's letter to the NYTimes was what - to prove Joe "stock options are not an expense" Lieberman flunked Finance 101? We already knew that!

Cent21 said...

Given that we've got the real, indisputable, current gross federal debt growing by nearly $600 Billion a year, and that debt is going to leave the on-budget portion of the tax code in need of a large tax increase within the next 15 or so years, and the size of THAT tax increase will presumably have a deadweight loss that is more than proportionate with the increase in taxes.

And this coincides with expressions that the Social Security trust fund holds only meaningless IOUs.

The only logical solvency reform in the short to medium term is to resolve the fiscal gap in the rest of government.

That would minimize debt service later, when real interest rates are likely to be rising because much of the trust fund debt will have to be rolled over to the public.

Presumably, we'd be better off running a more or less balanced budget starting within the next couple years, rather than needing to make much larger adjustments at a higher cost later.

If we increase the current Social Security surplusses now, it seems likely congress will simply spend the money coming in, creating a bigger hole to dig out of later.

An appropriate glide path for any necessary benefit cuts or payroll tax increases to deal with Social Security's infinite timeline seems to be about 15 to 25 years from now. This gives at least a couple more President's a shot at those reforms if President Bush fails, presuming Bush doesn't also fail with his primary current responsibility of managing the short term budget.

Andrew Samwick said...


Your statement is correct:

So if we pass now or we pass something in 2014 that applies to all born after 1950, there is NO difference AT ALL.

But I don't think it is realistic to posit that benefits can be cut as much or that taxes can be raised as much on this cohort if we wait 10 years to do it.

Patrick Sullivan said...

I think discussion in these terms is irrelevant. What is relevant is the way Robt. Samuelson is talking about it:

"In 1935 Americans 65 and older were 6 percent of the population. They're now 12 percent and by 2030 are projected to be 20 percent.

"....the real issue is not Social Security's 'solvency'. It is the total cost to the government of baby boomers' retirement, including Social Security, Medicare and Medicaid (which covers much nursing home care).

"The real issue is preventing those costs from becoming economically oppressive and politically poisonous. Even if the Social Security trust fund is made permanently 'solvent' -- in the sense that taxes cover benefits -- the costs of all federal retirement programs may still become undesirably high. In 2004 Social Security, Medicare and Medicaid were 8 percent of national income. Left alone, they'll reach 14.5 percent by 2030...."

It seems unrealistic to expect future generations of workers will acquiesce to paying that much of their income to take care of complete strangers. Even if productivity increases enough to allow them to have much higher real incomes.

Dan W said...

Through all the debates on saving Social Security, no one is discussing the question: "Why are we saving Social Security?" The fund is a program established for a purpose. The purpose was served and it is time to move on.

I am 33. Like many my age, I do not expect to ever receive a benefit from Social Security. By the time I am 70, the age will have moved to 90. By the time Im 90 (assuming) the age will be up to 100. Allow me put my contribution into a 401k in exchange for never drawing benefits.

The problem with the plans being floated are they do nothing to solve the root of the problem: "Too many people on Social Security." The plan was never intended for people to live on it for 20 to 35 years. At the plan's inception, people weren't expected to live on it fo 20 to 35 months. We neeed to reduce the number of people drawing benefits.

I'll be the first to volunteer to drop from the program. Let me deposit 5% of my payroll tax into my 401k and you can keep the other 7+% to fund the program. Make this mandatory for all people under 35. Make it optional for everyone else. Suddenly the program doesn't have an infinite horizon.

By allowing the SSA to issue bonds to cover the baby boomer hurdle, indexing the age to receive benefits to the life expectancy, and indexing benefits to something that is more realistic, the plan can remain solvent til it phases itself out. As the baby boomers begin to die off, the payroll taxes can be used to retire the bond debt.