Last month the New York Times printed an op-ed piece (“Social Security: It’s Worse than You Think”) by professors Gary King of Harvard and Samir Soneji of Dartmouth. Their piece asserted that the Social Security actuaries’ methods for projecting mortality are “antiquated,” prone to “interference from political appointees,” and result in projections that underestimate the Social Security financing shortfall. The piece was accompanied by a graphic showing certain “crazy” demographic projections purportedly arising under current methods, for example that “everyone who happens to be 55–59 in 2028 would die.”
The piece created quite a stir among those who write about Social Security finances. Several experts and other pundits published commentary on the professors’ criticisms. Among those most worth reading are rebuttals by Alicia Munnell, Paul Van de Water and Kathy Ruffing.
Many people have asked me for my reactions to the Times piece, I suspect for three reasons. First, as a public trustee I’m in a position to know whether the allegations hold water. Second, like the piece’s authors I’m well known to be on the pessimistic side concerning Social Security’s financial future, at least relative to other experts. And third, I’m the Republican public trustee during the Obama Administration, and so should be expected to be the first to complain if political appointees were interfering with the development of objective analytical methods.
Despite my own concerns about Social Security finances, I did not find the allegations printed in the Times to be persuasive. Below I’ll present some reasons as to why, in addition to some basic background information about the trustees’ projection process.
The trustees bear final responsibility for the assumptions and projections. The King/Soneji article criticizes the Office of the Chief Actuary for developing demographic projections with methods that are too “complicated,” “subjective,” and lacking necessary “expertise.” In fact the actuary’s methods are reviewed by quadrennial technical panels appointed by the Social Security Advisory Board and are subject to annual independent full scope audits. Its recommended assumptions and methods reflect ongoing consultation with a diverse array of outside experts. Ultimately the actuary’s office can only make recommendations for the demographic and economic assumptions; the trustees decide which ones to adopt. This involves a thorough annual review of the recommendations – scrutinizing them, questioning them and if the trustees so decide, modifying them. This is the process that has been sanctioned by lawmakers, in the text of the Social Security Act, for decades. Obviously the actuaries’ opinions carry great weight in the trustees’ deliberations, and the Chief Actuary’s statement of actuarial opinion is an important validating feature of the annual trustees’ reports. Still, neither I nor the other trustees put our names to the projections unless we are persuaded that they have been developed according to the highest standards of objectivity and analytical rigor.
The trustees’ report already accounts for much greater potential variability in possible financial outcomes than discussed by King and Soneji. The professors’ case that Social Security’s condition is “worse than you think” is based on their estimates that current projections are $800 billion off through 2031 and the program’s combined trust funds will be depleted in that year, in contrast with the trustees’ intermediate projection of 2033. By the standards of long-range forecasting, however, this is much smaller than a substantial difference. 2031 is well within the range of insolvency dates portrayed as realistic in the trustees’ uncertainty analysis, with the 10th percentile of possible outcomes being in 2030 and the 90th percentile in 2037. And while an $800 billion difference may sound large to the layman, all dollar figures involved over decades of Social Security financing are huge: for example, our report shows the calculated present value of the long-range imbalance ranging from $3.6 trillion (at the 10th percentile of possibilities) to $13.5 trillion (at the 90th percentile). The trustees’ report even warns that “readers should understand that the true range of uncertainty is likely to be larger than indicated.” The King/Soneji estimate isn’t evidence of a grave projection error, but rather fits securely within the bounds of reasonable variance that the trustees have long identified.
The “crazy” projections shown in the Times piece are not the trustees.’ More concerning is the fact that King/Soneji ascribe “crazy” projections, for example that all people aged 55–59 would be expected to die in 2028, to the Social Security actuaries. These are not Social Security actuary’s or trustees’ projections. The SSA actuary’s actual projections for future death rates can be found in a recently-published actuarial note; they bear scant resemblance to those presented in the Times piece. The “crazy” results in the Times op-ed were generated by adding the professors’ estimates of the effects of smoking and obesity to the SSA actuary’s model. Because the actuary’s model already incorporates changes in death rates arising from specific causes correlated with obesity and smoking, the authors’ additions effectively double-count the effects of these factors. The “crazy” numbers that result from this double-counting are of little relevance to the SSA actuary’s assumptions.
The allegations of political interference lack foundation. Perhaps the most disturbing aspect of the King-Soneji op-ed was the allegation that the demographic assumptions were prone to “interference from political appointees.” In my time serving as a public trustee I have not seen evidence of this. Dating back to the 1983 Social Security amendments, the process involves two public trustees, one of each political party, overseeing the development of the assumptions. In practice this usually involves a working group in which the public trustees engage directly with the staff of SSA, Treasury, Labor and HHS; the recommendations of this working group are then presented to the full board of trustees for possible modification and final approval. The assumptions initially recommended by the SSA actuaries are developed independently by career staff, not political staff. These assumptions are then subject to scrutiny by staff of the various departments (some of whom include political appointees) and by the trustees themselves, as envisioned under the Social Security Act. I have witnessed and sometimes participated in constructive disagreements about issues of presentation, emphasis and analysis, but I have not seen evidence of political considerations prompting suggested changes in the economic and demographic assumptions. Even if there had been, the public trustees would be there to critically review any such suggested changes. This highlights the importance of public trustees always being nominated and confirmed in a timely manner, to oversee the projection process. But the suggestion that the SSA actuaries’ recommended assumptions are influenced by political considerations has no basis of which I am aware.
King and Soneji are right to call for prompt legislative action to correct Social Security finances. In their piece they state that “the longer we ignore the (Social Security) problem, the more disruptive any change will need to be to keep Social Security alive.” This is true whether the projected insolvency date is 2031, 2033, or some other date. Changes required to restore sustainable solvency, even if enacted today, would have a significant impact upon workers and future beneficiaries, assuming that we don’t want to cut benefits for those now receiving them. By the early 2030s it would probably be too late to preserve Social Security self-financing; by then we would face a choice between dramatically cutting benefits for those already on the rolls, and politically implausible tax increases and/or enormous benefit reductions for subsequent recipients. For that matter, the King-Soneji mortality projections could well be right, just as the SSA actuary’s could; no one can say today whose demographic projections will ultimately prove to be most accurate.
The trustees’ projection track record is strong. While as Niels Bohr supposedly said, “Prediction is very difficult, especially about the future,” the trustees’ track record in this particular area is quite strong. At the time of the last major Social Security reforms in 1983, the trustees projected that unisex life expectancy at age 65 would hit 18.8 years in 2010. The actual figure turned out to be 18.7. Long before I became a public trustee, I wrote and spoke about the trustees’ impressive projection history. There have always been many willing to criticize the trustees’ assumptions, but given the inherent difficulty of long-range forecasting the trustees’ track record is remarkably good.
In sum, I do not believe that the future longevity gains predicted by professors King and Soneji, even if they prove correct, would produce a substantial difference in Social Security’s financial condition. Their expectations for Social Security finances are well within the zone of reasonable variance that the trustees have long identified. Moreover, neither the political interference nor the “crazy” results that they allege appear to have a basis in the trustees’ work. While I yield to no one in being concerned about Social Security’s financial future, I do not find additional reasons for concern in the recent Times op-ed.
Charles Blahous is a research fellow with the Hoover Institution, a senior research fellow with the Mercatus Center, and the author of Social Security: The Unfinished Work