Yesterday the Congressional Budget Office (CBO) corrected an error that it made in projecting the share of earnings that will be replaced by Social Security for those nearing retirement. In a report published last fall, CBO projected that for people born in the 1960s, the annual Social Security benefit for those retiring at age 65, would be 60 percent of their earnings for middle income retirees and 95 percent of earnings for those in the bottom quintile. The correction showed that benefits would replace 41 percent of earnings for middle income retirees and 60 percent of earnings for those in the bottom quintile. 

This mattered a great deal because the originally published numbers were quickly seized upon by those advocating cuts in Social Security benefits. For example, Andrew Biggs, who served in the Social Security Administration under President George W. Bush, used the projections as a basis for a column in the Wall Street Journal with the headline “new evidence on the phony retirement income crisis.” The piece argued that benefits were overly generous and should be cut back, at least for better off retirees. (To his credit, Biggs quickly retracted the piece after CBO acknowledged the mistake.) 

While this was a serious error, unfortunately it was not the first time that CBO had made a major error in an authoritative publication. In 2010, in its annual long-term budget projections it grossly overstated the negative effect on the economy of budget deficits. The 2010 long-term projections showed a modest increase in future deficits relative to the 2009 projections, yet the impact on the economy was far worse. 

The 2010 projections showed a drop in GDP of almost 18 percent by 2025, compared to a balanced budget scenario. This was more than twice as large as the impact shown in the prior year’s projections. The sharp projected drop in GDP could have been used to emphasize the urgency of deficit reduction. As was the case with the recent Social Security projections, CBO corrected its numbers after the error was exposed.

These two errors are troubling not only because of their size and importance to major public policy issues, but also because they should have been easily detected. Just to be clear, it is inconceivable that anyone at CBO deliberately put erroneous numbers in these publications. Hundreds of policy experts review these numbers after their publication. No one could think that faked numbers would escape detection.

Nonetheless, these mistaken numbers somehow got out the door at CBO. All CBO publications undergo internal review processes. They don’t just allow a staffer to type a document and post it on the web. In the case of the budget projections, anyone who was knowledgeable about CBO models should have immediately recognized the sharp break with prior projections, just as we did. Yet somehow no one reviewing the projections at CBO caught the mistake.

The same story applies to the projections of Social Security replacement rates. These numbers were far out of line with past projections as well as projections done by the Social Security Administration. Furthermore, the actual benefit numbers were included directly in the same publication in another table (Exhibit 9). They showed that the average benefit for a worker in the bottom quintile would be $11,000 a year (in 2015 dollars). Could this be 95 percent of these workers’ pay? The average benefit for a worker in the middle quintile was projected at $22,000. Is this 60 percent of the annual earnings of a middle income worker? 

Someone at CBO should have caught these numbers before they went out the door. They weren’t off by just a little bit, they were absurd. But somehow a number of economists and budget experts at CBO looked at these numbers and said they looked fine. The fact that both errors were in a direction that would tend to support cuts to Social Security is especially troubling. Would a major error in the opposite direction also escape detection?

This question is especially important in light of CBO’s latest set of budget projections. These projections show the annual deficit rising from a relatively modest 2.9 percent of GDP this year to 4.9 percent of GDP in 2026. CBO highlighted this projected rise in the deficit in the very first paragraph of the summary. Needless to say, the Washington deficit hawks quickly jumped on the CBO numbers to renew their calls for deficit reduction, including cuts to Social Security and Medicare. 

However a closer look at the numbers showed that the only reason that the deficits are projected to increase is that CBO projects a sharp rise in interest rates over the next two years. Higher rates are projected to persist over the rest of the projection period. The projected rise in interest rates leads to higher interest payments, and therefore a large deficit.

This projection of higher interest rates is troubling because CBO has projected a sharp rise in interest rates every year since 2010. It has been wrong in each of the last six years as interest rates remained low. So far, it looks like it will again be wrong in 2017, as long-term rates have fallen since the end of 2015. Given this track record, it might be reasonable for CBO to re-examine its models. After all, a model that consistently over-projects interest rates is not a very good model. 

Of course, if CBO were to project that interest rates remain somewhere near current levels, then there would not be much of a story of rising deficits, and little ammunition for those seeking cuts in Social Security and Medicare. There are certainly plausible explanations for CBO’s decision to stick to its model that don’t depend on a desire to cut these benefits, but the major errors that got through door are not encouraging in this respect.