The Accuracy of CBO’s Budget Projections

Posted by
Doug Elmendorf
March 25, 2013

Following a recent hearing, we were asked by a Member of Congress: “Does CBO ever go back and review its estimates of the budgetary impact of legislation?” Here was our answer:

Yes, CBO routinely monitors the budgetary effects of enacted legislation to help improve projections of spending and receipts under current law, as well as to improve cost estimates for new legislative proposals.

However, it is often difficult or impossible to determine, even in retrospect, the incremental impact on the budget of a particular piece of legislation. CBO regularly prepares cost estimates for legislation when bills are reported by committees of the House of Representatives or the Senate. In some cases, such legislation is changed before enactment. Although CBO often provides updated cost estimates (especially for direct spending provisions) prior to the enactment of legislation, proposals are sometimes amended after cost estimates are prepared. Moreover, in many cases the actual costs or savings resulting from enacting legislation cannot be identified; they may be a small part of a large budget account or revenue stream, and there may be no way to know for certain what would have happened if the legislation was not enacted. In fact, most of the cost estimates that CBO completes are for legislative proposals that are not enacted, so it is not possible to determine their accuracy.

A Regular Review Process

Because it is often not possible to determine how close the impact of a particular piece of legislation is to CBO’s initial projections, it is hard to make a general statement about the accuracy of our estimates. Nonetheless, CBO analysts undertake a detailed review of Treasury-reported outlays and receipts after the end of each fiscal year to learn as much as possible about how accurate the agency’s projections have been compared with both the original cost estimates for individual pieces of legislation and the current-law baseline projections (which reflect all legislation previously enacted). In addition, CBO updates its baseline projections a few times each year, and during those exercises, the agency carefully tracks and reports on changes from the previous baseline by separately categorizing and explaining changes derived from legislation, economic revisions, and other (technical) adjustments.

That annual process is useful in helping CBO prepare better projections going forward, even though it is sometimes not possible to discern exactly how much of a given year’s estimating error for a given program is directly attributable to a specific piece of legislation. A few examples of cases in which it is possible to match up results with earlier projections for specific pieces of legislation are summarized below.

Medicare Part D

The prescription drug program known as Medicare Part D is a relatively rare example in which actual spending can be directly compared to the projections contained in the CBO cost estimate. In most cases, legislation modifies existing programs; it is often not possible after enactment of such legislation to determine how spending for a modified program has changed specifically as a result of that legislation, or how much of future spending would have occurred even without the change in law. In contrast, the legislation that created Part D established a new component of Medicare with a system of new benefit payments, associated administrative costs, and payments from premiums and states.

The actual net cost of Medicare Part D has been much lower than CBO originally projected. For example, in its 2003 cost estimate for the legislation creating the program, CBO projected that Part D costs through 2013 would be $552 billion (the Administration’s estimate at that time was higher), whereas the agency now estimates those costs will total $358 billion—including actual net outlays of $304 billion recorded through fiscal year 2012 and the agency’s current estimate of $54 billion for 2013. The roughly 35 percent difference between the initial projection and actual results recorded thus far arises largely because CBO observed that recent growth rates for drug spending had been higher than the long-term trend, and we assumed that growth would remain above the long-term trend for most of the 10-year period following the creation of Part D. However, that growth rate dropped below its prior long-term average even before the new program was implemented in 2006—probably because patents expired for a substantial number of brand-name drugs (so consumption of those drugs was switched to lower-priced generic versions) and relatively few new brand-name drugs were introduced. In addition, enrollment in Part D has been lower than what CBO initially projected.

Recovery Act Spending

The American Recovery and Reinvestment Act of 2009 (ARRA) provided funding for a broad range of new and existing federal programs and reduced revenues through changes in federal tax law. Most of ARRA’s effects on federal spending and revenues have now occurred, and they have been roughly in line with the original estimates prepared by CBO and the staff of the Joint Committee on Taxation (JCT) at the time the legislation was considered by the Congress in early 2009.

CBO has closely monitored actual spending under ARRA for the past four years to help determine where the agency’s estimates of outlays and the timing of such outlays were too high or too low—both in total and for individual years and programs. Through fiscal year 2012, the outlays resulting from ARRA totaled $555 billion, about $36 billion (or 6.5 percent) above CBO’s original estimate of $519 billion for the 2009–2012 period. (Additional spending will occur over the next several years. In addition, JCT estimated that ARRA would reduce federal revenues by about $210 billion over 10 years, with most of that impact falling in 2009 and 2010.) The modest underestimate in spending under ARRA is accounted for by provisions related to unemployment insurance, nutrition assistance, and refundable tax credits; those costs were boosted by the weaker-than-expected economic recovery. Other spending from ARRA was a bit below CBO’s original estimate.

Some estimates were particularly close to the recorded results for each of the four years following the enactment of ARRA. For example, transportation spending under the legislation totaled $37.3 billion through 2012—only $0.5 billion (or a little more than 1 percent) below CBO’s original estimate. The results recorded for transportation in each of the years following enactment were very close to CBO’s estimates of the timing of those outlays.

Spectrum Auction Receipts

Legislation enacted in the past 20 years directed the Federal Communications Commission (FCC) to use competitive bidding (auctions) for licenses to use the electromagnetic spectrum when more than one party seeks such licenses. Spectrum auctions under such legislation have generated more than $50 billion in net offsetting receipts to the Treasury since 1994.

CBO’s estimates of spectrum auction proceeds under legislation enacted over the past two decades have sometimes been too high and sometimes too low. When estimating the budgetary impact of the Balanced Budget Act of 1997, for example, CBO projected that FCC auctions would generate about $25 billion in proceeds. Actual collections resulting from that legislation were about one-third less than projected. CBO also estimated spectrum receipts of about $25 billion from the auctions authorized by the Deficit Reduction Act of 2005, but the agency underestimated receipts for that legislation: Collections resulting from the 2005 act have been about 30 percent higher than the estimate.

Spectrum values fluctuate for several reasons, including changes in technology, market conditions, and the financial and strategic interests of individual wireless companies. Projections of receipts also reflect uncertainty about the quantity of spectrum that will be available for auction. CBO’s estimates attempt to reflect those uncertainties by representing the middle of the range of most likely outcomes.

Unemployment Insurance Spending

In 2008, lawmakers enacted the Emergency Unemployment Compensation program (EUC), which has been altered numerous times over the past several years. Under current law, that program expires at the end of December 2013. Adding together its estimates for the 12 laws that enacted and subsequently expanded and extended EUC, CBO estimated that benefits under that program would total $199 billion through the end of December 2012. According to the Department of Labor, the actual cost of EUC benefits has been $208 billion through December 2012, a difference of less than 5 percent.