New paper: “Do Social Security Surpluses Pay Down Publicly Held Debt? Evidence from Budget Data”

The Social Science Research Network has posted online a new research paper by Randy Mariger of the Treasury Department titled "Do Social Security Surpluses Pay Down Publicly Held Debt? Evidence from Budget Data." This article relates to recent econometric work begun by Kent Smetters and expanded by Barry Bosworth and Gary Burtless of the Brookings Institution and by John Shoven of Stanford and Sita Nataraj of Occidental College. All three research projects concluded that, by and large, surpluses in Social Security aren't truly saved, meaning that they don't translate into an improved overall budget balance. Rather, surpluses within Social Security tend to be offset by larger deficits elsewhere in the budget. (I summarized this research here.)

Mariger continues in this area, but argues that it is difficult to come to hard conclusions from budget data:

Being fair to future generations requires that Social Security be reformed in a manner that effectively prefunds a significant share of future Social Security benefit payments. All serious reform plans have this property. Prefunding is attempted exclusively in the Social Security trust fund in some plans, and it is attempted partly in personal retirement accounts in others.

Many analysts believe that Social Security surpluses are offset all or in part by lower non-Social Security surpluses. If the offset is 100 percent, then running Social Security surpluses does not increase the government's capacity to pay future Social Security benefits. In this case, reforms that rely on trust fund accumulations to make Social Security fair to future generations do so at the expense of a non-Social Security policy that is less fair to future generations.

The evidence on whether or not trust fund accumulations pay down federal debt is of two general types: formal statistical analyses of historical budget data, and informal observations of budget politics. Mariger (2008) reviews the recent history of budget politics and concludes that there is a substantial probability that Social Security surpluses are in large part offset by smaller non-Social Security surpluses. To complement that study, this paper attempts to draw out statistical evidence from budget data.

It is concluded that the budget data is essentially silent on the question of whether Social Security surpluses are truly saved. The reason is that the regression model specification is necessarily approximate, Social Security surpluses show little independent year to year variation, there are only 37 years of data, and spurious correlations mask the true relationships.

Mariger does appear to agree with the above researchers, however, "that there is a substantial probability that Social Security surpluses are in large part offset by smaller non-Social Security surpluses."