Answers to Questions About Dynamic Analysis

Posted by
Keith Hall
July 1, 2015

In May, the Congress adopted a concurrent resolution on the budget for fiscal year 2016. That resolution requires CBO, to the greatest extent practicable, to incorporate macroeconomic effects into its 10-year cost estimates for major legislation that Congressional committees approve. Such estimates must also include, when practicable, a qualitative assessment of the budgetary effects for the following 20 years. Incorporating such macroeconomic feedback into cost estimates is often called dynamic scoring.

Last month I gave a presentation at the Heritage Foundation to describe how CBO will implement dynamic scoring in response to that new requirement. Since the Congressional budget process was established in the 1970s, CBO’s cost estimates have typically not included dynamic analysis. (Among the few exceptions was our cost estimate in 2013 for immigration legislation that would have substantially increased the U.S. labor force.)

Apart from cost estimates, CBO has produced estimates of how some proposals that would significantly change federal spending and tax policies would affect the overall economy, as well as how such effects would affect the federal budget. Most recently, in June, CBO published an estimate of the budgetary effects of repealing the Affordable Care Act, which analyzed the main budgetary and economic effects of repealing that law. That report, in particular, serves as a good example of how CBO intends to present such dynamic estimates under the new requirement. (You can see these reports and other work in this area on our Dynamic Analysis web page.)

In light of our recent work and the budget resolution’s new requirement, I want to share CBO’s answers to several questions about dynamic analysis that we received from Members of Congress after hearings earlier this year before the Senate Budget Committee. (You can view the complete set of questions and CBO’s answers from the January 28 and June 16 hearings on our website.)

1. How will CBO present estimates that include macroeconomic effects?

CBO anticipates that the form in which the agency provides this information to the Congress will evolve over time depending on what sort of presentation seems most useful. Such cost estimates will include all of the information that typically would be included if macroeconomic effects were not incorporated in the analysis, as well as additional information related to the macroeconomic effects themselves and the uncertainty surrounding those effects. For example, if applicable, CBO would provide separate estimates for on-budget and off-budget effects, including macroeconomic effects to the extent possible. (Off-budget effects include changes in Social Security spending and revenues as well as in spending by the U.S. Postal Service.)

CBO expects that the specificity of the assessments that the agency will provide about the effects of legislation in later decades will vary depending upon factors such as the amount of time CBO has to conduct the analysis, the complexity of the legislation being considered, the capability of the tools that CBO has to assess the legislation’s effects, and the agency’s judgment about the uncertainty of the analysis. A qualitative assessment of effects in later decades might look something like this:

  • Including macroeconomic effects, CBO expects that the legislation would (increase/decrease) the deficit during the 2015–2025 period and would (increase/decrease) it in the following 10 years; those (increases/decreases) in the following 10 years would probably be (larger/smaller) than the ones projected for the 2015–2025 period.

In some situations, CBO might conclude that it has sufficient basis to provide some quantitative information about a subsequent decade, perhaps along the lines of one of the following:

  • Including macroeconomic effects, the legislation would (increase/reduce) cumulative deficits by about 0.X percent of GDP over the decade following the 2015–2025 period.
  • Including macroeconomic effects, the legislation would (increase/reduce) cumulative deficits by about $X billion over the decade following the 2015–2025 period.

2. How accurate do you think dynamic scoring is compared to the current scoring methods employed at CBO?

Conventional analysis of proposed legislation incorporates analysis of behavioral responses by individuals and businesses to incentives in the legislation but does not incorporate effects on the economy as a whole, meaning that potential macroeconomic effects exert no feedback on the budget. Sometimes, however, that macroeconomic feedback may be important, so that trying to account for such effects may increase the likelihood of producing an accurate answer. To accomplish that result, CBO has devoted considerable effort during the past several years to improving its ability to estimate fiscal policies’ macroeconomic effects and the associated impact on the budget. That effort has included extensive reviews of relevant economic research and yielded tools that the agency can apply to help analyze a variety of policies.

To present and explain its approach, CBO has published several reports and working papers:

CBO’s conventional analyses and analyses that include macroeconomic effects both try to give a single point estimate for each of the next 10 years of how proposed legislation would affect the federal budget deficit. How accurate those estimates are can be measured by the degree to which a set of point estimates of the effects of various changes in law differ from the eventual actual outcomes. (For additional discussion about how to measure accuracy, see CBO’s Economic Forecasting Record: 2015 Update.) Such accuracy is determined by two properties of the set of cost estimates, centeredness and spread:

  • Centeredness indicates how close the average of the estimates is to the average of the eventual actual values.
  • Spread reflects the dispersion of the differences between the estimates and their eventual actual values around the average of those differences.

Estimates that are more centered and have less spread are more accurate. One concrete measure of this concept of accuracy is the probability that the actual cost will eventually fall within a certain range around the reported estimate.

Hence, under certain conditions, including feedback effects would probably make estimates more accurate than estimates based only on conventional analysis:

  • If the estimates are for policies that are likely to have significant feedback effects,
  • If the estimates of those feedback effects are centered around the actual effects, and
  • If the spread of those estimates is small.

Often, however, estimates that include significant feedback effects will be more centered around the actual outcomes (because they aim to account for all of the effects) but will also have considerably larger spread because of the inherent uncertainties involved in estimating macroeconomic effects.

That combination may or may not be more accurate, as CBO defines the term, but CBO expects that incorporating effects from macroeconomic feedback will improve the quality of estimates for complex legislation with many types of provisions and significant budgetary effects. Incorporating those effects makes the estimates more centered, and errors in estimating the effects stemming from particular provisions tend to cancel each other out. For that reason, CBO emphasizes the importance of having estimates be in the middle of the distribution of potential outcomes.

Determining the accuracy of CBO’s cost estimates—conventional estimates and estimates including macroeconomic feedback—for legislation that is ultimately enacted is often quite difficult and sometimes impossible. In general, the actual costs or savings resulting from enacting legislation are a small part of a large budget account or revenue stream and cannot be separately identified. Nonetheless, to improve its approach to estimating, CBO regularly scrutinizes errors in its projections, reviews data on actual spending patterns for federal programs, and consults outside experts on those programs.

3. If we use dynamic scoring principles for the tax code, shouldn’t we also use this scoring for federal investments—such as education, research, and infrastructure—as well?

The budget resolution specifies that CBO will incorporate the budgetary impact of macroeconomic effects into cost estimates for legislation involving both mandatory spending and revenues with effects exceeding certain thresholds, but not for appropriation bills or bills that just authorize appropriations. CBO also will incorporate the budgetary impact of macroeconomic effects in cost estimates for legislation when requested to do so by one of the Chairmen of the budget committees.

Appropriation bills have funded most federal investment, and CBO has begun to develop tools to estimate the macroeconomic effects of changes in federal investment so funded. Last year’s analysis of the President’s budget proposal included estimates of the budgetary impact of such effects, as did CBO’s analysis of one set of alternative policies in the long-term budget outlook. A forthcoming report will provide additional information.

Although the budget resolution does not require it, CBO could, upon request by a committee Chair or Ranking Member and given sufficient time, prepare analyses of the macroeconomic effects of proposed changes in funding for federal investment; such analyses would be separate from any cost estimate associated with the request. In those analyses, CBO would estimate how changes in federal investments affect the economy in relation to the amount of federal investment already incorporated into CBO’s economic forecast, which are the amounts implicit in CBO’s baseline projections. For many types of investments, those amounts—through 2021—are shares of the discretionary funding amounts specified under the Budget Control Act of 2011, as amended. (For projections involving total nondefense discretionary funding, CBO’s estimates incorporate the assumption that investment would equal its average share of such funding from 1980 to 2008—a period when that share was roughly constant.) In the following years, they are similar amounts with adjustments to account for the effects of inflation. Thus, only incremental changes to federal investment that differ from the projected amounts would be estimated to have macroeconomic effects in relation to CBO’s baseline projections of the economy.

Investments that affect the economy might include, for example, infrastructure, research and development, and education. How an increase in federal spending on such investment would affect the economy—and hence the budget—would differ in the short term and long term:

  • In the short run, as with other government purchases, increased federal investment spending is estimated to boost output by increasing total demand for goods and services. That boost tends to be larger when output is well below its maximum sustainable amount and the Federal Reserve’s response to changes in fiscal policies is likely to be limited.
  • In the long run, increases in federal investment spending—if not offset by decreases in other spending or increases in taxes—are estimated to have opposing effects on economic output. Increased federal spending on investment generally raises productivity in various ways, which tends to boost output. However, higher spending also causes the government to borrow more, which tends to “crowd out” private investment, lowering output.

Keith Hall is CBO’s Director.