Trustees v. CBO: Projection Accuracy for Social Security

Both the Social Security Trustees (Trustees) and the Congressional Budget Office (CBO) produce annual short- and long-term projections of the financial status of the Social Security retirement, survivors and disability insurance programs. The purpose of these projections is to ascertain the security of indicated benefits, estimate the cost to the plan sponsor, and to evaluate the impact of any changes in the plan/program. The Trustees have produced an annual Report for more than eighty years, as directed by federal law, while CBO has done a similar projection exercise for the last twenty years, as part of its assessment of the federal budget. Heretofore, however, legislative proposals have typically been brought to the Social Security actuary and not to CBO for scores.

Both organizations devote considerable government resources to the projection task, coming to somewhat different results, and so it is reasonable to ask which does a better job. One might think that the Trustees, given their long track record, experience, and close access to real-time data on the operations of the Social Security programs, would have the advantage. On the other hand, given CBO’s tighter focus on the current and likely future state of the economy and the federal budget, and with a large team of analysts and modelers using more sophisticated methods, perhaps CBO has an advantage. Ultimately, it is an empirical question.

Here I measure the performance of both organizations’ short-term projections, going back five and three years, of revenues, costs, and net cash flow compared to actual values, for the years 2011 through 2022. Short-term relative performance from these recent years may be considered more relevant to the administration, Congress, and the public for budget purposes and as the Trust Fund exhaustion date nears.

In Figure 1, I compare in percentage terms the five-year ahead projections of the Trustees and CBO versus the actual non-interest income (revenues) coming into the Social Security Trust Funds. Clearly, the Trustees are consistently too optimistic, while CBO was initially too optimistic (neither organization anticipated the Great Recession), but then was slightly too pessimistic. Overall, CBO has a better track record here, as measured by the mean squared error.

Figure 1. Percentage Difference in Non-Interest Income, 5-Year Projection vs Actual, 2011-2021 (% of Actual)

In Figure 2, we see the same comparison for the three-year ahead projections – both organizations are clearly more accurate closer in, although the Trustees are still consistently too optimistic, and, with the exception of a big negative error for 2022, CBO is more accurate. Figures 3 and 4 give the same comparisons for the cost of the programs, that is, outlays for benefits and administrative expenses. Both organizations are fairly accurate even five years ahead, although the Trustees over-projected somewhat in recent years. Again, CBO has an edge over the Trustees.

Figure 2. Percentage Difference in Non-Interest Income, 3-Year Projection vs Actual, 2011-2021 (% of Actual)

Figure 3. Percentage Difference in Cost, 5-Year Projection vs Actual, 2011-2021 (% of Actual)

Figure 4. Percentage Difference in Cost, 3-Year Projection vs Actual, 2011-2021 (% of Actual)

Finally, in Figure 5, I compare, in billion dollar terms, the five-year (a) and three-year (b) ahead projections, respectively, of the net cash flow (non-interest income less cost) compared to actuals. Visually, neither organization seems superior to the other, with both being too optimistic in the early years and CBO too pessimistic in more recent years. In terms of mean squared error, however, the Trustees outperform CBO.

Figure 5. Difference in Cash Flow (Projected minus Actual), 5-Year Projection (a) and 3-Year Projection (b) vs Actual, 2011-2021 (billions USD)

What are the implications of this empirical analysis? Neither organization is clearly superior in its projection accuracy. But given the confusion and extra resources that two sets of projections cause, the general experience that “two heads are better than one,” that each organization has some unique qualifications, focus and experience, and my strong belief that a unified projection is superior to multiple siloed projections, I recommend that the two organizations combine their efforts to produce one set of Social Security projections. This way CBO would have an official role in scoring legislative proposals for Social Security reform—a natural task given its authority and prominence on the short- and long-run budget. As an independent agency of Congress, CBO could replace the bipartisan Public Trustees, whose positions have been empty for nearly a decade, with no prospect for replacement, to balance out the Trustees from the administration and their staffs. CBO also has a better track record than the Trustees in terms of the timeliness of its publications according to schedule.

---

Note: All calculations are the author’s based on data from the annual Social Security Trustees Reports and the Congressional Budget Office website.