The Schizophrenia of Social Security Reporting

by Bruce Webb

When I first started paying serious attention to Social Security reporting back in 1997 I was struck by its curious mix of precision and fluidity and how that mix moved from the latter to the former as you went from the data tables to the top line numbers and then on to whatever coverage the Report received by the media. That is by the time the numbers hit the New York Times everything was in declarative mode “The Trustees project the Trust Fund WILL go to Depletion in year X unless taxes are increased by 2.xy%”. But when you examine the actual Report in detail you find that ‘year X’ and ‘2.xy%’ are simply mid-points of a probability distribution. Worse when you examine the Reports on a year over year basis you find that that mid-point changes in ways that can whip the long-term probability tail significantly. Yet in the reporting there is always the unstated assumption that THIS year’s data set is closer to projecting the next 75 years than last year’s set. And there are pragmatic reasons for that, for one thing we have that extra year of historical data to work with and for another you have to start with something but the downside is that it installs a certain sense of fatalism, that the numbers are what they are and that we are all more or less helpless victims of them. Well the truth is a little different, there are reasons why the mid-point number falls where it does each year and why it changes year over year, and some of those reasons are subject to intervention, we can and do make changes in the present that move those future numbers.

So between changes in policy and variations in the economic inputs the uncertainty going forward simply gets wider and wider both within OAS and DI projections and between them, by the time you get thirty and forty years out you are talking very wide probability bands. Yet the rules of the game are set in a way that we are more or less forced to work with the 75 year projection. Which is frankly a little silly particularly in the case of Social Security where the danger of a fix that is too big is actually greater than one which is too small, something I discussed in number 10 of the original Angry Bear SS series The Danger of Low Cost

The first step towards curing this schizophrenia is in recognizing that it is mostly induced. The economists who fuel the studies that drive the apparent need for Social Security ‘reform’ are fully aware of the uncertainties but choose to ignore them because the longer time frames allow them to use scarier numbers to urge more drastic solutions. But we don’t have to be bound by that, instead we can sit down, examine the models and determine pragmatically when they begin to diverge enough to call for different initial policy choices. If I had to pick a number it would be fifteen years out. By that point each of the three models deployed by the Trustees has reached its ultimate numbers when expressed in percentage form, the Office of the Actuary not deceiving themselves into thinking they can model precise changes to things like productivity on a year to year basis after even year eight or nine. And the Reports give us enough numbers that we could choose 2025 as our planning horizon, but since after ten years the data is only reported at five year intervals going fifteen years out from 2011 or 2012 gets more problematic. On the other hand the Trustees do supply a convenient moving 25 year number, and one where there is already enough variation between the top and bottom of the models to justify drawing the line for planning purposes. See Table IV.B4 below the fold.

I think it is apparent why I would prefer fifteen year planning windows, already by 2034 the distribution is from -1.86% of payroll for High Cost to +1.12% for Low Cost. Also worth noting is that OAS in isolation is under Intermediate Cost in almost perfect balance over 25 years at +0.05%, additional evidence that a solid short to medium term plan of ‘Nothing’ is still ideal for OAS. On the other hand even under the most optimistic model entertained by the Trustees DI barely would inch into positive territory, the chance of a 0.30% fix now sending it into overfunded territory before we could rebalance things being pretty remote indeed.

But in any event it is pretty crucial to get analysts and reporters to grasp that these models are not static and that successive Report years don’t necessarily get us closer to a full understanding, instead the inherent uncertainty of future economic developments will always swamp refinements of the methodology. While we can expect the Office of the Chief Actuary to do their best, we have to keep in mind that every pronouncement about the future outlook of Social Security needs to have a whole set of “if and only if”‘s attached to it and the best we can do is to lob solutions somewhere near the center of the spread.