Friday, January 21, 2005 and Social Security's Unfunded Obligations

FactCheck.Org weighs in today with an article in response to the President and Vice President both making statements about Social Security's unfunded obligations, "Does Social Security Really Face an $11 Trillion Deficit?" As I discussed in my first post on Social Security, the present value of the excess of projected benefits over projected revenues and the balance in the Trust Fund was $10.4 trillion as reported in the 2004 Trustees Report. This is the number in question.

As is typical with articles by, this one first identifies whether the statement is literally true and then tries to point out what is missing:

Both are correct -- but fail to mention that nearly two-thirds of that colossal bill doesn't come due until after the year 2078.

The table from the Trustees Report also shows that the unfunded obligations are projected to be $3.7 trillion considering only the revenues collected and benefits paid through 2078, the last year of the traditional 75-year projection period.

The rest of the article focuses on a disagreement between the recommendations of the 2003 Technical Panel on Assumptions and Methods that reported to the Social Security Advisory Board and a letter by the American Academy of Actuaries on whether it is appropriate to use a measure of the program's unfunded obligations that extends forever.

The position of the Technical Panel was that such measures be used as a way to provide additional information about the program's actuarial balance and that it also be expressed as a percentage of taxable payroll over that time period. As shown in the table from the Trustees Report noted above and as discussed in my earlier post, the $10.4 trillion corresponds to 3.5 percent of taxable payroll. It would be more informative if the public discussion noted that fixing the projected shortfall in the system on a permanent basis would require a change in finances equivalent to raising the payroll tax rate from 12.4 to 15.9 percent, immediately and forever. The article says this clearly, and I agree with that assessment.

The AAA letter takes an opposite position, arguing that the infinite-horizon estimates should not be used in general. I am frankly quite puzzled by the AAA's argument. Consider this passage:

With regard to the infinite-time-period estimates, the Committee begins its analysis by noting that the results of the 75-year statutory valuation are themselves subject to extreme uncertainty. Consider the situation of actuaries or economists in the year 1928 attempting to project demographic and economic parameters 75 years into the future - to 2003. They likely would have missed the Great Depression, World War II, the baby boom, the influx of women into the labor force, etc. Nobody, no matter how intelligent or educated, could have anticipated these very significant events.

Nobody would have asked these experts to anticipate these events. Nor is their inability to foresee these major events a compelling reason to deliberately ignore the years after the 75th when making projections. What we would have asked is for these experts to give us the most comprehensive measure of the system's future costs and revenues, along with their assessment of the uncertainty in that mesure, so that policy could be made in the most informed way possible. I would like to know how exactly the AAA suggests that informed policy can be made with an arbitrary cutoff point.

The letter continues with:

Although methods of demographic, economic and actuarial analysis have improved greatly over the years, the sources of error in past valuations of OASDI have been unforeseen — and really unforeseeable — large-scale changes in the U.S. society and economy. We see no reason to believe that similar, unforseeable large-scale changes will not occur in the future. Given the uncertainty of projections 75 years into the future, extending the projections into the infinite future can only increase the uncertainty, rendering the results of limited value to policymakers, unless an infinite projection is used exclusively to identify an ultimate reversal of an apparent trend demonstrated in the 75-year projection.
This last statement is not true. "Extending the projections into the infinite future" certainly will "increase the uncertainty" in those projections, but that is not the only thing that will occur. Extending the horizon for the projection will also shift the central estimate of the system's revenue, because it includes additional years that are very likely to be worse than the average year during the projection period. Again, how can the AAA possibly argue that the trajectory of the cost rates in the last years of the projection period is irrelevant to an assessment of the program's financial health?

Returning to the larger issue, here are some guidelines that I would recommend for the discussion of Social Security's financial condition:
  1. Do not rely exclusively on measures of solvency like the projected date of trust fund exhaustion or the actuarial balance over an arbitrary period like 75 years. In addition to these measures, discuss the projected imbalance in the system over the infinite horizon and express it in a sensible metric, like the 3.5 percent of taxable payroll.
  2. Recognize that there is uncertainty in all projections and that this uncertainty is greater for more distant years. The Trustees Report has made some progress in this respect (see this Appendix), but this does not yet include years beyond the 75th and the language used to describe this exercise needs to be refined so that it is clear.
  3. Expect that reality will differ from projections, but do not let that be an excuse to leave the system as written in current law persist with large projected gaps. At all points in time, current law should have the system in balance, based on the best projections available.
  4. Reinforce point #3 with accounting systems that do not enable the off-budget surpluses in entitlement programs to facilitate greater deficits in the on-budget account. This means that we should not use the unified budget deficit as a metric for deficit reduction and focus instead on the (larger) on-budget deficit.
And, as I have noted elsewhere, point #3 should apply broadly to the government's financial activities, including the General Fund and Medicare.

Other blogs commenting on this post


Anonymous said...

I really appreciate the sense you frequently bring to this subject of Social Security pension funding Andrew. Thanks.

I especially appreciate the recommended guidelines for these discussions.

Kirk from Colorado

JG said...

I remain completely puzzled about why everybody keeps talking about the "solvency" of Social Security in terms of the trust fund, and over terms such as 75-years and forever -- parsing it all to the nth degree -- instead of in term of the general fund, which of course must provide every penny financed by the trust fund, and which is going to be taking the big hit in the clearly visible future.

E.g.: The trust fund keeps SS solvent by providing financing for SS from 2018 until 2042. Fine.

But right in the middle of that period, 2030, the SS Trustees say the _general fund_ needs of SS will be so large as to require a 22% increase in income taxes from today's levels as a percentage of GDP, to cover trust the cost of trust fund operations.

And then there's the other trust fund that nobody mentions, for HI-Medicare. That's going to require another 13% increase in income taxes. That's a 35% increase in income taxes combined, as a % of GDP, just to cover the operations of the trust funds! And rising forever more after.

See Chart E:

Now I submit that when the day actually comes in the 2020s that taxpayers have to vote on raising their taxes by 35% just to cover the trust funds -- on top of another 28% by 2030 to fund Medicare proper (a 63% tax increase total!) -- precious darn few of them are going to be thinking "Thank God we have the trust fund to still finance Social Security for years to come!"

So why, oh why, is everybody fixated on parsing the discounted liabilities of SS more than 75 years out, with _nobody_ talking about the points of GDP that the trust funds will be addding to the income tax bill just 20 years from now? Isn't that a solvency issue?

Can someone explain this to me?

I'll take it even further. The last time Congress had to increase taxes to cover SS benfeits, in 1983, the political crash that resulted produced a significant re-working of the program which changed the benefit structure to cover the financing shortfall 50% with benefit cuts.

Logic dictates that the next crash comining before 2030, which will be much bigger than the 1983 one -- MUCH bigger -- will result in much bigger structural changes to the program than we saw in 1983. (Taxpayers are not going to just swallow a 63%-and-perpetually-rising increase in income taxes and go happily skipping along their way!).

With such changes coming it is just totally _nonsensical_ to parse SS liabilities and solvency over 75 years. Angels holding their Junion Prom on the head of a pin have more to do with reality.

So, when people talk about "solvency" of SS, why don't they talk about in terms of the actual cash demand it (and the HI trust fund) will place on general revenue in 2020, 2025, 2030... which will have the _real impact_ on the solvency of the government and the welfare of the living people involved?

Reall, I'm at a loss to understand it.

Anonymous said...

I note that a 15.9% tax rate would produce additional surpluses through about 2025-2030; however, by 2080, the last date in current SSA projections, the cost rate (intermediate case) is about 19.5%. How, then, can the 15.9% rate show solvency for the infinite horizon case? Somewhere along the line the "trust fund" will disappear and the cost will exceed an income rate of 15.9%

Andrew Samwick said...

With a 15.9 percent payroll tax rate, the accumulations to the Trust Fund are sufficient so that the interest on its balances each year is sufficient to cover the annual deficit when those deficits subsequently appear.

Whether it is a sensible policy depends on what will happen to the government's spending as a result of the additional years of off-budget surpluses. If the government decides to spend them rather than saving them, then they do nothing to enable future generations to actually cover those deficits. My own view is that these additional revenues will be spent, and that is a good enough reason to force the government to disgorge the money into personal accounts if additional revenues are brought in as part of reform.