1994-96 Advisory Council Report
Assumptions And Presumptions
Carolyn L. Weaver
Which Assumptions to Use?
One of the contentious issues the Council faced during its first two years of deliberations was what baseline set of economic and demographic assumptions to use in evaluating various proposals. Much debate, and certainly a considerable amount of time, centered on one question: should we use the Social Security Board of Trustees' intermediate assumptions, those considered "best guess" by the Social Security Administration (SSA) actuaries and generally relied on by policy makers, or should we modify these assumptions to incorporate a different CPI assumption? For a period of time, it was agreed that the Trustees' assumptions would be used without modification. Subsequently, a decision was made to reduce the CPI assumption by 0.5 percent in recognition of the fact that the Bureau of Labor Statistics was expected to modify the way it measured the CPI--and to use the modified assumptions for purposes of "scoring" the various plans. Subsequent to this, a decision was made to reduce the CPI assumption by 0.3 percent. Before the Council managed to resolve the issue--and well before the Council would conclude its business--the Bureau of Labor Statistics had announced its measurement changes, and the decision was made to conform to these changes. The CPI assumption in the 1995 Trustees= Report was reduced by 0.21 percent.
Simultaneously, but implicitly, a decision was made to assume that this reduction fed directly through to increase all of the Trustees' assumptions pertaining to real variables. This seemingly minor technical adjustment had the effect of increasing the ultimate real wage growth assumption and other key economic assumptions. (In one fell swoop, the productivity of labor over the next 75 years was assumed to be higher than previously assumed.) This decision regarding the presumed effect of a CPI measurement change on real economic variables was made without the Council having a basic understanding of the analytical issues involved.
Quite possibly, during our first two years of deliberations, more time was spent haggling over the CPI assumption--and having the actuaries generate new financial projections on our plans with each new twist--than on any other matter. The reason, of course, was that the change in the CPI assumption (together with the presumed changes in real variables) reduced the size of the measured financing problem--by about one-seventh on a long-term average basis--and allowed the Council to propose smaller adjustments in taxes and benefits than otherwise would have been needed. This was important to members seeking to minimize the appearance of a serious long-range financing problem (none of whom supported a direct reduction in the cost-of-living adjustment).
Unfortunately, in the quest to have a "common baseline" for comparing the three plans, much more time was devoted to the question of adjusting the CPI assumption than to the question of the reliability of the intermediate projections as a basis for policy development--or the potential value of the actuaries' high- and low-cost projections. While the Technical Panel on Assumptions and Methods, convened by the Advisory Council, stressed the uncertainties inherent in the Trustees' long-range projections--and proposed new analytical techniques for recognizing these uncertainties--the Council itself proceeded to elevate these projections to a position that cannot, in my view, be justified.
It is extremely important to appreciate the limits of the actuaries' projections. The assumption about economic growth, for example, is invariant with respect to whether the long-range deficit is eliminated by an immediate increase in the payroll tax of 14 percent (2.17 percentage points); a gradual series of rate increases ultimately totaling 50 percent (or about 6 percentage points); or a 16 percent reduction in long-range benefits. This and other economic assumptions are invariant with respect to whether social security is advance funded to a very large degree, with investment in real private capital, or pay-as-you-go financed. As a result, any potential effects of accumulating large personal savings accounts on, say, national saving, capital accumulation, labor supply and retirement decisions, and economic growth are simply ignored. These economic effects would surely swamp the effects of the policy details that consumed the Council's time, such as whether, given a fixed set of assumptions, to use a 35- or 38-year period for computing average lifetime earnings, or whether a percentage amount in the benefit formula should be 13 percent, 12.4 percent, or 9 percent. These latter policies affect, in known ways, little more than a hypothetical person's benefit level based on a hypothetical wage profile.
The Illusion of Precision
Insisting that every reform package, and every single item proposed--no matter how small (such as increasing the earnings computation period) or large (such as the means of financing a shift to fully funded individual accounts)--be reduced to a single set of actuarial numbers, based on a single set of assumptions (using a model(s) understood by few, if any, but the actuaries), was like trying to force square pegs into round holes. An illusion of precision was created that simply can not be justified and the range of options that could be considered by the Council was sharply limited. The truth is that many things simply do not fit neatly into the actuaries' round holes (some might say black box).
This is not to disparage the work of the Office of the Actuary, which was enormous in quantity and is generally regarded to be of high quality. (Some of the best policy analysts at SSA are actuaries in that office.) It is to disparage the Advisory Council's over-reliance on the actuaries' projections, which can easily, and erroneously, be taken by the public and the media to imply that the most important effects of proposed reforms are known, can be quantified with precision, and pertain to social security tax and benefit levels; that the aggregate economic effects of the various reform plans are identical or quantitatively unimportant; and that the probability that each plan will eliminate the long-range funding gap is identical. But this is nonsense. The most important effects of these plans--on aggregate economic activity and on individual well-being now and in future decades--can not be quantified with precision; they can only be predicted or inferred based on rigorous empirical studies grounded solidly in economic theory. And such effects surely would be found to differ dramatically given the very different way in which these plans distribute taxes, benefits, and debts. While the SSA actuaries are well aware of this, the public would have little reason to question the implications that flow from the Council's rigid reliance on the actuaries' projections.
Narrowing Our Field of Vision
More importantly, perhaps, reliance on the actuaries' projections narrowed the range of options that could be considered and the way in which these options could be evaluated. The problem was two-fold: First, because researchers generally do not have access to SSA's data and models, there were only a few people in the land who could generate revenue and cost estimates for the Advisory Council, namely, a handful of actuaries in SSA's Office of the Actuary, and no one that could proceed to model, in any reasonably timely manner, the likely economic effects of proposed changes. Second, options that fell outside the aegis or expertise of the SSA actuaries tended to fall into a bureaucratic never-never land. Readers will note the absence of any substantive policy analysis of the Council's proposed changes in pension policy, which presumably required the expertise of several offices and agencies within the Labor and Treasury Departments--let alone outside researchers.
When the options under consideration were boiler-plate, such as raising the retirement age faster or higher, or raising the payroll tax by various amounts, there generally was no lack of ready numbers, even if there was a serious lack of economic analysis. When the options under consideration amounted to true reform, as in the case of personal accounts in our PSA plan, there was a serious shortage of useful information of any kind forthcoming from SSA. This created a natural bias toward the familiar--tinkering changes in taxes and benefits--and a high hurdle to jump for those of us interested in more fundamental reforms.
The idea of moving toward a system of personal accounts was squarely on the table a year and a half ago, in the summer of 1995, and there were numerous options that could and should have been explored. Allowing people the right to voluntarily opt into a system of personal accounts, paying off outstanding liabilities in entirely different ways, and structuring the tier one benefit as, say, a minimum-benefit guarantee for low-wage workers (e.g., as a supplement to the accumulations of low-wage workers) or a means-tested benefit for low-income retirees, financed out of general revenues, are among these options. The list goes on. The Advisory Council was provided with no issue papers or research materials with which to explore the options and trade-offs in any concrete way. Once a proposal to privatize half the retirement program was found to have some general support, all of the effort went into having financial estimates prepared for this and the other two options under consideration--with some Council members spending countless hours fiddling with technical details to make sure that all of the numbers "added up," not to mention attempting to influence the choice of every possible assumption not contained in the Trustees' report, such as the administrative costs of various investment options and the method of calculating money's worth ratios. This effort consumed the final year (or more) of our deliberations.
The ultimate irony is that, in the end, proponents of one plan, the Maintenance of Benefits (MB) plan, backed away from the single largest component of their plan--centralized investment of 40 percent of trust fund assets in private equity. Conveniently, it was too late to ask SSA to generate an entirely new set of financial estimates for the plan, so it is evaluated as if Social Security had the benefit of a higher rate of return on trust fund assets and thus could finance higher benefits (or maintain lower taxes) than is actually the case.
A Final Thought
The three reform options presented in this report embody three very different visions of what Social Security is and can be in the decades ahead. These visions reflect our different assessments of the long-term viability of the current pay-as-you-go system, our different views on the abilities of working men and women to manage their financial affairs, and our different understandings of the workings of modern financial markets. Sadly, because of the time and resources consumed in "scoring" plans and ensuring technical "consistency," these visions--and the range of reform options that logically flow from them--may be lost in the technical and actuarial detail of this report. If Congress is to reform Social Security in the next few years in a way that promotes national saving and economic growth and truly enhances the value of the system for younger workers and future generations, the debate must begin. That debate can not be allowed to be side-stepped, superseded, or derailed by the presence or absence of a set of actuarial projections.