For the last four years I have had the honor of serving as one of Social Security's two public trustees (along with former CBO Director Robert D. Reischauer), co-authoring the annual trustees' reports. These reports are required under the Social Security Act and serve as the primary resource on Social Security's current and projected financial condition. Long before my own service as a trustee I was an avid reader of these annual reports, coming to regard them as a vital information source that is, in the main, well serving the public.

It is inevitable and appropriate that the trustees' annual projections are publicly debated. An unusually rigorous and thoughtful critique was recently published by Konstantin Kashin, Gary King and Samir Soneji, in one form in the Journal of Economic Perspectives (JEP) and another in Political Analysis. Barron's also published a widely-circulated article on this work. These articles offer a number of criticisms of the trustees' projection record as well as the processes followed by the Social Security chief actuary's office.

As a currently-serving trustee I have been asked for my view of the Kashin-King-Soneji work. Summarizing very roughly, their factual observations and analyses strike me as essentially correct, though I disagree with many of their interpretative conclusions (full disclosure: I was one of many sources interviewed by the authors in the course of their extensive research). In this piece I will review some of their critiques pertaining to the trustees' projection history. In a follow-up piece I will turn to their criticisms and recommendations with respect to the process by which the projections are developed.

One quick note at the outset: the articles frame criticisms primarily with reference to the methodologies of the Social Security Chief Actuary (OCACT). Formally, however, all of the assumptions and methodologies employed in the annual trustees' reports are decided upon by the trustees themselves, with the chief actuary simply signing an opinion certifying that "the techniques and methodologies used are generally accepted within the actuarial profession and that the assumptions and cost estimates used are reasonable." In practice, of course, the actuary's recommendations are highly influential upon the trustees. It is the norm for the trustees to accept most such recommendations unless there is a consensus among them to direct otherwise, which does happen on occasion. Ultimately the trustees are responsible for the assumptions made and the projections that flow from them.

The following paraphrases (in my own words) the Kashin-King-Soneji criticisms of the trustees' projection record, and offers my own perspectives on the issues raised.

Criticism: Until 2000, Social Security projections were mostly unbiased, with errors falling on both sides of the line with roughly equal frequency. Since then the projections have become increasingly inaccurate in overstating Social Security's financial health.

As a purely factual observation, this is undeniably true. The main reason is predictably simple: the Great Recession caused Social Security finances to be much worse than the trustees anticipated.

A good quick way to see this is to examine the annual projections for the combined Social Security trust fund ratio (the ratio of trust fund assets to one year's expenditures, x 1 00). As the graph shows, the big divergence between projections and reality developed after 2008 (the recession hit in December 2007). Thus, any projection window including the effects of the recession will generally show much larger projection errors than periods excluding the recession. (This graph like others in this article uses fiscal years to allow for comparisons with other forecasts.) Even before the recession there was some projection inaccuracy as always, but the qualitative divergence between the projection lines and the eventual reality clearly took place in 2008.

The main way in which the Great Recession threw the prior projections off was by depressing Social Security revenues (mostly payroll taxes, but also benefit taxation and interest income), as shown in the next graph. (The graph uses current dollars to mirror the presentation in the trustees' reports, though I normally prefer to adjust for price inflation or GDP growth.The reader may note that the recession made most everything smaller than previously projected, including price inflation). Technical nuances aside, it is readily apparent that the post-2008 deterioration relative to projections was primarily a product of lagging revenues.

By contrast, the trustees' spending projections were pretty much spot on, both before the recession and right on through it.

So the post-2000 trustees' projections were off primarily because they failed to predict the Great Recession. I can't fault the previous trustees for this because none of the other major forecasters did so either. The only other government entity producing independent Social Security trust fund forecasts during this period was the Congressional Budget Office (CBO). Before the recession, as typified in the example shown in the graph, CBO's projections were further off than the trustees.'

This is not to criticize CBO, which fully deserves its reputation as a high-quality and unbiased forecaster of government finances. It's to make the points that the trustees' biggest projection errors were largely due to the Great Recession, that they would have been subjected to widespread ridicule had their projections anticipated a recession of such severity, and that they actually got closer to the mark than their forecasting contemporaries.

In fact, throughout this period the trustees were subjected to intense and ill-founded goading to be more optimistic in their projections. A fashionable meme arose on the left side of the political spectrum that the trustees' projections were consistently too pessimistic about program finances because they were badly underestimating economic growth. There never was a plausible basis for this myth, as I showed in my 2007 paper, "Have the Social Security Trustees Been Too Conservative?"

The Great Recession rendered that line of criticism ludicrous (though, remarkably, not completely extinct) by placing Social Security finances on a far worse trajectory than previously anticipated. The trustees of the period deserve credit for withstanding this pressure, and for achieving the modest distinct ion of being less wrong than most others when the Great Recession made everyone's projections look bad.

Criticism: The projections have consistently underestimated life expectancy by amounts exceeding the range of expressed uncertainty. This shows overconfidence in the projections and has led to an underestimation of system costs. The actuary/trustees have also ignored useful advice from technical panels that would have mitigated these errors.

Kashin-King-Soneji present a compelling case that the trustees' reports have consistently underestimated life expectancy for the last decade and a half. And they are further correct that key aspects of life expectancy -- such as life expectancy at age 65 -- are now longer than projected even in the illustrative "high cost" scenarios in earlier trustees' reports. It is also true that several recent technical panels have opined that the projections were understating life expectancy. The Social Security actuary's office acknowledges much of this in a recent memo explaining some of the sources of mortality projection error.

All this granted, a lay-reader of the Kashin et al work risks coming away with an exaggerated picture of these factors' importance. Nearly half of the JEP paper focuses on life expectancy projection shortcomings. But as the same paper shows (see its Figure 4), only a lesser fraction of the trustees' cost projection errors are attributable to these factors. Longevity's role among cost underestimation factors is actually even less than that, because the total projection error is the net of some factors contributing to cost underestimation and some contributing to cost overestimation. But most importantly, total forecasting errors in recent years were much more about lagging revenues than they were about cost projections. In the grand scheme of things, the longevity projection issues were small potatoes relative to the dominant factors contributing to projection error.

It risks being somewhat selective to highlight the trustees' divergence from the technical panels on longevity estimates when there were other equally significant disagreements where the trustees' judgments appear so far to have been the superior ones. During the time period covered by Kashin et al, there have been three technical panels. Each of the three has recommended that the trustees increase their longevity assumptions. Each of the three has also recommended that the trustees increase their wage growth assumptions. The table below shows how each of these recommendations would have affected the actuarial balance expressed as a percentage of the tax base.

This table actually understates the point because the wage growth assumption changes would almost certainly have had bigger effects during the near-term period studied by Kashin et al than those pertaining to mortality. In any event the bottom line should be clear: the trustees' projections indeed suffered because they did not adopt some outside recommendations. But broad acceptance of such recommendations would not have improved the projection record. If the trustees are to be criticized for the recommendations they wrongly rejected, they should be praised for the ones they rightly rejected.

The test of a forecasting record is not whether one can find isolated errors via hindsight but rather how it compares to other forecasts made on the basis of the same information. Forecasts made before the Great Recession generally look pretty bad to us now. But within that context the trustees' projection record, while imperfect, is highly respectable among forecasters.

Part two of this series will explore the Kashin et al critiques with respect to the trustees' projection process.

Charles Blahous is a senior research fellow for the Mercatus Center, a research fellow for the Hoover Institution, a public trustee for Social Security and Medicare, and a contributor to e21.