What fresh hell is THIS?
When I first started studying Social Security in detail sometime late in 1997 I made what to me was kind of an amazing discovery. Social Security ‘crisis’ then and now tended to be perceived and discussed within the deterministic frame of Boomer Retirement, kind of a ‘demography makes destiny’ thing. Every Boomer who would ever exist was already on the face of the planet and we had reasonably good data about longevity improvements. Which seem to leave us with an open and shut case: more Boomer retirees living longer vs a declining pool of workers equals crisis down the road. Thus it made some sense to talk and think about Social Security in fixed terms of Boomers becoming eligible for retirement at known points in known numbers resulting in Social Security going ‘broke’ (however defined) at specific points in the future. The result is that reporting on Social Security tended to use deterministic language like ‘will’ as in “The Trustees tell us the Trust Fund will run short in 2034”. (And for those who point out ‘I thought the date was 2041’, well that is the point of this piece.)
But when you actually examine the numbers it became clear that in fact they were doubly contingent with any given Annual Report projecting a range of potential outcomes (your ‘intra-contingency”) while successive Reports could and did have a different range (your ‘inter-contingency’). In each case the objective seemingly fixed ‘will’ starts blurring into the subjunctive ‘would (assuming contingencies A, B, and C)’.
Well what does this gobbly-goop mean in real terms? Well lets start from the following table from EPI. Changes in Trustees Projections Over Time If we just look at the first two columns for 1996 and 1997 we are presented with just what we would expect from our deterministic model: dates of Trust Fund depletion remaining set between years with the cost of the fix going up. Open and shut case for crisis? Well not so fast.
First we can examine the Cost of Inactivity on an inter-termporal basis. Not doing anything in 1996 meant the cost of the fix went up from an immediate hike in FICA of 2.19% to fix SS to an increase of 2.23% or a change of .03% of payroll. But the flip side of that is that not doing anything had the effect of LEAVING 2.19% of 1996 your payroll dollar in your and your employer’s pocket in 1996 for that year. If we simply ignore the effects of current utility of that dollar, of current year interest, and of current year inflation we can calculate that it would take 2.19/.03 or 73 years before ‘Nothing’ became a nominally bad bet for 1996. If we add in the effects of utility, interest and inflation we can see that each makes ‘Nothing’ just that much more a good bet. Let’s say that you made $30,000 in 1996. 2.19% of that is $657 which given the employee/employer split means $328.50 in take home pay (more if you take the classical position that the real incidence of the employer match falls on the employee). It doesn’t take a huge rate of real return to make a diversion of that $328 into some other investment pay off; calculated in terms of risk/reward a strategy of ‘Nothing’ made sense in 1996. That is in retrospect. On the other hand it would have taken a real deterioration in outlook and a pretty big boost in payroll gap to make ‘Nothing’ a losing bet in real terms.
If we examine the remaining columns in the table we can see that ‘Nothing’ turned out to be a winning bet both in nominal and real terms each year from 1997 to 2004, in each year ‘Nothing’ left money in your pocket while the Cost of Inactivity actually went down, ‘crisis’ steadily got pushed out in time and shrunken in magnitude. Now certainly you wouldn’t ‘know’ this in advance, on an inter-temporal basis the contingency is kind of out of your control and critics might just claim you got lucky. Well maybe, but in actuality the numbers fell out as they did. And there was pretty good reason to expect that they would.
Why? Well it is the intra-temporal contingency. As it turns out the Social Security Report does not project a single outcome, instead each provides a range of outcomes with their best guess giving a mid-point Intermediate Cost alternative with a more optimistic Low Cost marking one end of the probability range and a more pessimistic High Cost marking the other. Now one can, and I do, dispute that the Office of the Chief Actuary has actually properly presented the range and so argue that Intermediate Cost is in fact too pessimistic, but for our narrow purposes today we can lay that aside. Instead we can just examine the first year numbers in terms of the Cost of Inactivity. If we take our strategy of ‘Nothing’ and actual numbers come in in line with Intermediate Cost we can expect a positive Cost of Inactivity which we can then if we choose discount for assumed real interest. On the other hand if the numbers come in better than Intermediate Cost the Cost of Inactivity shrinks even before any discounting. And if the numbers happen to come in near the top of the range the Cost of Inactivity actually goes negative, meaning dollars left in your pocket with no downside at all.
Now getting back to the numbers. When I read the 1997 Report I noticed two things. One was that the 1996 Low Cost projection resulted in a straight out fix of Social Security, doing ‘Nothing’ would produce a fully-funded Social Security system with no changes in payroll tax, benefits or retirement age. And what kind of economic performance would be needed? As it runs out only a future that more or less mirrored 1996. If the future on average looked like the immediate past we would be home free. But we did not need to make a bet on Low Cost long term, we did not need a crystal ball to calculate the Cost of Inactivity for any given year, instead we only need to look at the current year numbers in the newspaper. If 1997 underperformed 1996 but came in better than Intermediate Cost we could expect the Cost of Inactivity to shrink and possibly go negative (i.e. leave dollars behind and a smaller projected gap going forward). By the time I saw the 1997 Report the year was mostly over and it didn’t take a lot of bravery to predict that year end would look better than IC projections. And so it proved, the 1998 Report showed a payroll gap down from 2.23% back to 2.19%. And ten more years of Nothing has brought the gap down to 1.70%.
Which is why I laugh in the faces of people who claim I am being naive in claiming that Social Security is Not Broke. They imagine that I am betting long on Low Cost when in reality I am betting short against Intermediate Cost. Now as it turns out it looks like 2008 will turn out to be a least a nominal loser for me, we will see with the release of the next Report in March. But I am still betting that the real Cost of Inactivity by adopting a strategy of Nothing will prove to be a winner over the medium term and probably over the long term, because while my bet may be short term nothing is keeping me from peeking at numbers for the out years and the Inter-temporal contingencies are pretty low hanging fruit.
In any event there is no ‘will’ associated with any projection whether that be short or long term. Each Report year presents us with new dynamic models with new intra-temporal and inter-temporal data sets which can only be properly evaluated using concepts like ‘if’ and ‘if and only if’. The standard deterministic ‘Baby Boomer demographics’ model just doesn’t capture the contingencies. And talking points that may have been reasonable enough in 1996 are stale, stale, stale today. A fact that seems to elude a certain subset of commentators here.