Social Security Actuarial Gap: the Time Series

Cross posted from The Angry Bear [1]

The 2010 Social Security Report is due out any day now and I thought I could lay some groundwork for a number-based discussion of its implications, [Ed. note: The report has been delayed until June 30] starting with changes to the 75-year projection over time.

1996-2.19%
1997-2.23%
1998-2.19%
1999-2.07%
2000-1.87%
2001-1.86%
2002-1.87%
2003-1.92%
2004-1.89%
2005-1.92%
2006-2.02%
2007-1.96%
2008-1.70%
2009-2.01%
2010-??

Each Report Year the Trustees give an estimate of how much payroll tax would have to go up immediately to deliver Long Term Actuarial Balance, meaning full payment of scheduled benefits and a Trust Fund with a one-year reserve at the end of the 75th year. In this post I want to explore the reasons why this number can go either up or down over the course of a single year, sometimes in a pretty dramatic manner, how can one Report Year whip the tail by as much as +/- 10%? Geekiness under the fold.

The first thing to understand is that the bias in this number by rights should be upwards for two reasons. Consider 1998, in point of fact we did nothing in the faces of a 2.23% actuarial gap from the previous year, meaning that extra dollars were not collected and so contributed neither principal or interest to the Trust Fund, all things being equal this should have added to the cost of an ultimate fix, because as 'everyone knows' the longer we wait the more any fix is going to cost. Well the numbers show that not to be true, in the face of inactivity the total cost can actually go down, why that is to be explored a little later. The second cause for an upward bias is the result of the change in valuation period, in 1998 what had been year 1, that is 1997 dropped off, and what had been year 76 became the new year 75 and so got included. And the Trustees tell us that change alone will add 0.06% to the gap. Meaning that if we were 100% confident in our modeling we would expect to see an uptick in that amount each and every year.

Well a single glance at this time/number series shows that we are not justified in that confidence, when people claim that the Trust Fund 'will' run dry in 2037, what they mean knowingly or not is 'if every bit of our methodology is correct and so too all of our economic and demographic assumptions about the future'. Well obviously we don't know what is going to be happening in 75 years, or even 10 years. For that matter only a handful of people not named Baker or Schiller were predicting this years economy even 3 years ago.

This uncertainty is fully recognized by the professionals in the SSA Office of the Chief Actuary, a fact that is shown in three ways. First the numbers above are drawn from is known as the Intermediate Cost alternative, the OACT also provides a more optimistic Low Cost alternative, and a more pessimistic High Cost alternative to present a range of possible outcomes. Second the OACT doesn't pretend to be able to predict future business cycles after the first ten years, instead all of their economic number series under all three alternatives settle out into what are called the Ultimate values, and finally the OACT shows its work in the form of Sensitivity analyses and Stochastic projections that show how those outcomes vary in response to changes to individual variables and in light of past variations, all of which are published in the Annual Report.

Which leads me to my first conclusion here: Infinite Future is Hooey. Given the probabilistic variation revealed even over the medium term and particularly given that an outcome that shows the system fully funded over the 75 year period (which is the result of Low Cost), and given the substantial uncertainties in even year over year projections as revealed above trying to influence policy by an 'Unfunded Liability' over the 'Infinite Future Horizon' is not just misguided, it is thoroughly dishonest. Particularly given the time series.

In 1997 Social Security could still reasonably be said to be in crisis, the 2.23% projected actuarial gap had been steadily increasing over the few years before just as you would expect from a crisis left unaddressed, the fact that the system had finally reached a Trust Fund ratio of 100 in 1993 and so met the short term test did not mean the whole program was out of the woods, as long as the actuarial gap was ticking upwards there was good reason to 'Save Social Security First', after all under then current projections the Trust Fund would go to zero in 2029.

Well a funny thing happened, starting in 1998 the gap, even in the face of inactivity started to drop, in the first year just a little but in 1999 and 2000 by a lot, 10% improvement per year in a 75 year outlook. And it was in response to this that Dean Baker and Mark Weisbrot published Social Security: the Phony Crisis [2] and I developed my own plan for Social Security which I cleverly dubbed "Nothing". In 2000 Greenspan was happily assuring us that via the miracle of the Great Moderation we had conquered the business cycle and we could enjoy 3% productivity and 4% Real GDP forever. Numbers that if correct were far better than the ones of Low Cost which showed Social Security vastly overfunded going forwards.

Which leads to a second conclusion. When it came to Social Security the Bush Administration was thoroughly dishonest from the start, they promoted their tax cuts based on a set of economic numbers that if they had actually come about would have saved Social Security along the way. Yet they willfully ignored the number series above and convened a Social Security Commission that still relied heavily on the rhetoric of crisis and 'we can't afford to wait'. Well the hell we couldn't, the ways the projected numbers were moving during Bush's first term there was positive evidence that a short term policy of "Nothing" was in fact optimal.

As noted at the outset the natural bias in this series in the face of inactivity is upwards due to theoretical taxes not collected in the previous year combined with the built in increase due to change in actuarial balance. Leaving aside the specific reasons for changes in projections from year to year the fact that four years of inactivity from the time Bush was elected in 2000 to his re-election only increased the actuarial gap by 0.02% and even at that was down by 0.03% from the previous year made the whole Social Security Tour of Spring 2005 itself throughly dishonest. These numbers show clearly why "There is No Crisis" was a relatively easy campaign to win, our side had the numbers.

Well except one. In 2003 a new measure of Social Security solvency suddenly appeared in the Reports, instead of measuring the payroll gap over the standard 75 year window, the Trustees would add a new number projecting Intermediate Cost over the Infinite Future Horizon. Now given the existing probability range over even the first 75 years and the rapid changes in outlook seen since 1997 this was more akin to witch magic than anything, but it produced a very useful number on its own. All of a sudden the President's Men could point away from a number that was under 2% and on balance shrinking and to a number that was almost twice as high at 3.5%. And they needed a number that high and badly because of something I pointed out in an adage coined in 1997:

"If privatization is necessary, it won't be possible. If privatization is possible, it won't be necessary."

For movement Republicans the real Social Security crisis was that it existed in the first place. But due to its general popularity a head-on assault was thought politically impossible, so in order to sell their ideologically preferable system of Private Accounts they were forced to play the solvency crisis card, and to present the better returns on equities as being the key to the solution. Unfortunately for them there were three problems here. One their models assumed traditional returns on stocks, which in turn required growth rates in line with historical trends, which rates if sustained were significantly better than those projected under Intermediate Cost assumptions. Hence the ditty above. A secondary problem is that along side their ideological commitment to Private Accounts was their even stronger commitment to tax cuts which had been sold on claims that such cuts would boost productivity. If tax cuts had actually worked as advertised then Social Security would have been saved from crisis along the way. A third problem was structural, any move from traditional Pay-Go Social Security to self-funded Private Accounts had to somehow handle the transition costs. And there was no way to do that at a cost to the worker that would not exceed the 1.89% payroll gap. Which would reasonably lead an informed worker to ask "Exactly what is in this for me that I should take this extra risk when I could just except the tax increase and guarantee a 100% benefit?"

And there was no answer to that in 2005, even the use of a 3.5% Infinite Future number didn't rescue the calculations, there was no numerically based challenge to the proven plan of "Nothing". To see why we have to look a little deeper at the numbers and what I call "The Cost of Inactivity".

In 2001 the process of actual shrinkage in actuarial gap stalled. There is some reason to believe that this stalling was deliberately engineered so as not to make the sales job for private accounts even harder than it was. People in a position to know on both sides of the issue insist that deliberate manipulation of the numbers in the Report is impossible, but both Prof. Rosser and an originally skeptical Arne took a hard look at the internals of the data tables and numerically if not practically the case is pretty solid. But either way a steady number as seen from 2001-2005 is an irrefutable argument for "Nothing" for two reasons. One the same performance numbers that drove down and then kept steady the actuarial gap also moved the date of Trust Fund exhaustion out in time, between 1997 and 2005 the date of depletion had been pushed out from 2029 to 2042. Not only had that date been moving at more than a year per year, suggesting that the event itself would never happen, it was being pushed out past the point of maximum Boomer demographic impact, a close look at the numbers show that the thirty or so years after mid-century will be relatively easy on Social Security as it handles a cohort of Gen-Xers that is numerically smaller than the Boomers that will have moved off the stage and the still working Millennials behind. In 2005 the case for simply waiting out 'Crisis' was excellent. Because every year of inactivity meant the functional equivalent of a tax cut in the amount of the fix not applied.

Well the case for "Nothing" got more problematic with the gyrating numbers from 2006-2009 and what is likely to be a disappointing number for 2010. Which is why I moved off "Nothing" to support the "NW Plan". But in any event it is important that all decision making on this be driven by the numbers and not by slogans. And that means digging into the data tables and not relying on the MSM to bring you the news. Hopefully this post will help you cut through the spin certain to be coming down the road in the next couple of weeks after Report release.