A Review of CBO’s Estimate of the Effects of the Recovery Act on SNAP

In 2009, the Recovery Act boosted monthly benefits for SNAP. The resulting increase in spending on SNAP benefits from 2009 to 2013 was greater than CBO had estimated. This report discusses that underestimate and the reasons for it.
Summary
In February 2009, in response to significant weakness in the economy, lawmakers enacted the American Recovery and Reinvestment Act of 2009. That legislation funded a broad range of new and existing federal programs and reduced revenues through changes in federal tax law. CBO monitored actual spending under the Recovery Act (or ARRA) for several years after that enactment to assess the accuracy of its estimates of outlays for the legislation—both in total and for individual years and programs.
For the 2009–2013 period, outlays resulting from ARRA totaled $596 billion, about $54 billion (or 10 percent) above CBO’s original estimate of $542 billion for that period. Provisions related to unemployment insurance, nutrition assistance, and refundable tax credits together account for the underestimate. In particular, CBO’s estimate of spending for the Supplemental Nutrition Assistance Program (SNAP, which helps people in low-income households to purchase food) accounted for a substantial share of that underestimate. For all other programs, taken together, CBO’s estimates of outlays from ARRA over the 2009–2013 period were within 6 percent of the actual total. This report focuses on the underestimate for SNAP and the reasons for it.
Among its many other provisions, ARRA increased monthly benefits for SNAP participants. When the law took effect, the maximum monthly benefit for a household of four was raised by 13.6 percent, from $588 to $668. Benefits for households of other sizes increased similarly, and the law specified that the maximum benefits would remain at those amounts until the routine inflation adjustment that otherwise would have been applied to the benefit amount exceeded the increase provided under ARRA.
At the time of the law’s enactment, CBO estimated that ARRA would increase spending on SNAP, relative to CBO’s baseline, by $20 billion—or by 3 percent of total projected SNAP spending—over the 2009–2019 period, with nearly all of that spending occurring in the first five years. CBO now estimates that ARRA increased total spending on SNAP by $43 billion over the 10-year projection period. That is, the ARRA-related increase in spending on SNAP was more than double the amount CBO estimated in 2009.
About $19 billion of the estimated $20 billion increase in SNAP spending was for benefit payments. The rest encompassed outlays for changes to administrative funding, the Food Distribution Program on Indian Reservations, block grants for Puerto Rico and American Samoa, and time limits for able-bodied adults without dependents. CBO underestimated ARRA-related outlays on SNAP benefits by $22 billion over the 2009–2013 period.
This report explains the methods that CBO used in February 2009 to estimate additional federal spending as a result of the provision in ARRA that raised the maximum SNAP benefit. It also provides details about the factors that contributed to CBO’s underestimate of that spending. In particular, CBO’s estimate of ARRA-related spending on SNAP benefits in February 2009 was substantially below the actual amount over the 2009–2013 period for two main reasons, both involving CBO’s baseline projections:
- About 90 percent of the difference is attributable to CBO’s baseline projections of SNAP’s average monthly benefits, which were substantially higher than the actual amounts. In its January 2009 baseline (which preceded the law’s enactment), CBO overestimated the routine inflation adjustment applied to SNAP benefits, which was based on the agency’s projection of the consumer price index (CPI) for food at home. That in turn caused CBO to underestimate the ARRA-related change in average benefits per person. Had CBO correctly projected the change in the CPI for food at home, the projection of ARRArelated spending on SNAP benefits would have been higher than it was and substantially closer to the actual amounts.
- The remaining 10 percent of the difference is explained by CBO’s underestimate of SNAP participation, which arose in part from CBO’s baseline projections of economic conditions.
Two factors affected actual ARRA-related spending on SNAP benefits between 2014 and 2019 that CBO could not have anticipated in completing its February 2009 estimate. The first was the 2010 enactment of legislation that ended ARRA-related spending on SNAP benefits after October 31, 2013. (In October 2010, CBO estimated the effects of repealing those benefits.) The second was the federal government’s shutdown in October 2013.
Any comparison of CBO’s estimates with actual results is complicated by legislation that is enacted after an estimate is completed. CBO does not attempt to predict future legislative changes or their effects on outlays when it prepares its baseline budget projections or its cost estimates. In this case, ARRA-related spending on SNAP between 2014 and 2019, and thus CBO’s underestimate of that spending, would have been substantially larger if not for the subsequent legislation that eliminated the increase in SNAP benefits provided under ARRA. Had the benefit increase not been repealed, CBO estimates, ARRA would have increased spending on SNAP benefits by another $21 billion over the 2014–2019 period, bringing the total estimated increase in outlays for benefits to $62 billion over the 2009–2019 period, or $43 billion more than CBO estimated in 2009.
CBO can compare its estimates of ARRA-related spending on SNAP with actual spending for two reasons. First, ARRA required federal agencies to report separately on spending arising from that law’s provisions. Typically, changes in spending caused by changes in legislation are reported as part of overall program spending and therefore are difficult to isolate. In this case, however, the Treasury has reported its estimates of actual ARRA-related spending. Second, SNAP benefits increased for all participating households in an identifiable way. Thus, the actual cost of the increase is easier to estimate than it would have been if the effects of the legislation depended more on participants’ behavior or on decisions made by state and local governments. For example, the effects of a policy that would change the application process would be much more difficult to estimate because of other factors that also influence the process.