Dynamic Scoring and Budget Estimations

Deborah Kobes, Jeff Rohaly

The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.

---

Q: Why is the House holding hearings about using "dynamic scoring" to evaluate the revenue impact of tax and spending proposals? Does this mean that current estimates are flawed?

A: Dynamic scoring is meant to provide a more complete picture of the budget effects of tax and spending proposals by incorporating the macroeconomic effects of the legislation. Official estimates already account for the microeconomic effects on individual behavior. Incorporating macroeconomic feedback effects to reflect how policies might affect economic growth would be, in theory, desirable. In practice, however, dynamic scoring would not improve the reliability of budget estimates. Because the process would necessarily require many subjective decisions, it would also risk opening up the scoring agencies to criticism of bias.

Discussion

In May, the Congressional Budget Office (CBO) testified before the House Budget Committee on federal budget estimation, and the Joint Committee on Taxation (JCT) testified at another hearing held by the Ways and Means Committee. These testimonies examine the implications of changing the estimation process of the two agencies, responsible for scoring spending (CBO) and revenue (JCT) proposals. The current official estimates now incorporate the microeconomic behavioral responses to proposals. For example, reductions in marginal tax rates are assumed to alter the composition of employee compensation away from fringe benefits and into taxable wages and salaries. Similarly, increases in gasoline excise taxes are assumed to reduce the consumer demand for gasoline and thus are estimated to bring in less additional revenue than a purely static score would suggest.

Over the last several years, each agency has expanded its analysis of how proposed and enacted legislation would affect the macroeconomy including effects on labor supply, national saving, and growth (which reflects the interplay between the labor supply and national saving effects). The CBO has released these studies as independent reports or appendices to official estimates. In those studies, CBO is not constrained by the rules governing baseline revenue projections. It can therefore provide a range of possible outcomes based on differing assumptions about the macro effects of policy and possible future responses to policy by Congress and the Federal Reserve. JCT has developed a macroeconomic equilibrium growth model (MEG) that, in concert with its microsimulation models, estimates the macro effects of tax legislation. JCT testified that it is still in the process of refining its estimation techniques in this area.

The recent hearings consider whether these companion macro-level analyses should be incorporated into official dynamic scoring forecasts by the CBO and JCT. Proponents of dynamic scoring argue that this process would give a more complete picture of the cost or benefit of federal proposals. For example, to the extent that tax cuts or government investment in infrastructure spur growth, they lose less revenue or cost less than current methods would suggest. In addition, providing lawmakers with the effect of tax and spending legislation on the macroeconomy could persuade Congress to adopt more sound economic policies. Those in favor of dynamic scoring also argue that ignoring macroeconomic effects means that the current system of scoring budget proposals produces estimates that are certainly incorrect since every nontrivial proposal has at least some macroeconomic impact.

Unfortunately, measures of macroeconomic feedback effects are very sensitive to assumptions that are subjective. Several sources of uncertainty are particularly difficult to resolve:

• Future fiscal policy: A dynamic scoring system would require considering not only a current proposal, but also the future policies that it would induce, and the economic responses to those subsequent policies. This requires assumptions about not only the economic but also political climate. For example, the macro effects of a tax cut will differ depending on whether it is financed with reduced government spending or borrowing (which would necessitate tax increases in the future); there would be no objective way of determining which method of financing would be adopted.
• Monetary policy response: To the extent that dynamic scoring included short-run macroeconomic responses, it would require considering monetary policy in addition to fiscal policy, because the Federal Reserve Board could react to changes in the economy due to new legislation. Policies of the Fed could either offset or accommodate the macroeconomic effects of tax and spending legislation in the short run.
• Expectations of taxpayers: For provisions that are scheduled to expire, the scoring agency would need to make a determination of whether taxpayers believe the sunset would actually occur or whether the legislation would be extended. The economic effect of a tax cut will often differ if taxpayers perceive it as transitory rather than permanent.
• Business cycle: Predictions of business cycles and economic fluctuations are extremely uncertain and varied. CBO does not currently attempt to forecast downturns or recoveries beyond an 18-month to two-year period. Assumptions about these cycles, however, greatly alter the potential macroeconomic effects of tax and spending legislation. For example, the short-term effects of a policy depend on whether the economy is below production capacity rather than at or near full employment. Although the CBO is working to include the business cycle into its predictions to a greater degree than currently, the impact of the changing economy could cause inconsistent estimates.

Each of these macroeconomic issues requires a series of assumptions with no standardized methodology for determining them. Dynamic analysis subject to these macroeconomic uncertainties could be helpful descriptively, even if only to show how sensitive a proposal would be to various changes in these assumptions. However, producing an estimate in the form of a single revenue or cost number would be misleading. The uncertainty surrounding critical assumptions, furthermore, may expose the estimating agencies to attacks of bias from those politically invested in a proposal.

Dynamic scoring would also consume scare resources within the CBO and JCT and would require a large degree of interaction among the agencies, especially in the case of large proposals that incorporate many revenue and spending measures. Given that many estimates are prepared under intense time pressure, adding a dynamic component could delay legislation or add to the likelihood of errors. Revenue measures estimated by JCT to have a macro impact would affect the cost of spending proposals, which would then need to be incorporated into estimates by CBO, and vice versa. Although proponents of dynamic scoring insist that only major proposals that are certain to have a substantial effect on national saving or labor supply or which would have large demand-side effects during an economic downturn would need to be dynamically scored, an objective rule would need to be established as to when a proposal qualified for dynamic scoring. This could create unfortunate incentives: for example, supporters of a tax cut would have an incentive to propose a large enough reduction to qualify for dynamic scoring. Given that current scoring methods already tend to make proposals look more affordable than they are - by ignoring debt-service costs, for example - such incentives may exacerbate the long-run budgetary deficits.

Additional distortions would be created if only tax, and not spending, measures were subject to dynamic scoring. This would create an incentive to formulate proposals in the form of tax measures, through refundable tax credits for example, even if the proposal would be simpler and more efficient as an expenditure program.

It is generally agreed that lawmakers should have access to the macro effects of tax and spending legislation. But, as both the CBO and JCT testified, this should come in the form of supplementary information to the traditional scoring estimates of proposals. This would allow the agencies to provide a range of possible impacts of legislation, accompanied by the assumptions used to produce the estimates and the sensitivity of the results to changes in those assumptions. Given the degree of uncertainty inherent in current methods of macroeconomic forecasting, true dynamic scoring would not allow the consistent and comparative cost estimates currently provided by the CBO and JCT.

---

Further Reading: Congressional Budget Office. 2002. "Federal Budget Estimating." Testimony before the U.S. House of Representatives, Committee on the Budget. Washington, DC: May 2, 2002.

Gale, William. 2002. "Issues in Budget Reform." Testimony before the U.S House of Representatives, Committee on the Budget. Washington, DC: May 2, 2002.

Joint Committee on Taxation 2002. "Written Testimony Concerning Modeling the Economic Effects of Changes in Tax Policy." Testimony before the U.S. House of Representatives, Committee on Ways and Means, Subcommittee on Oversight. Washington, DC: May 6, 2002.

Orszag, Peter. 2002. "Macroeconomic Implications of Federal Budget Proposals and the Scoring Process." Testimony before the U.S. House of Representatives, House Rules Committee, Subcommittee on Legislative and Budget Process. Washington, DC: May 2, 2002.

Penner, Rudy. 2002. "Role of the Congressional Budget Office." Testimony before the U.S. House of Representatives, Committee on the Budget. Washington, DC: May 2, 2002.