Statement of the Hon. R. Glenn Hubbard, Chairman,
Council of Economic Advisers

Testimony Before the Subcommittee on Oversight
of the House Committee on Ways and Means

Hearing on the Modeling the Economic Effect of Changes in Tax Policy

May 7, 2002

Chairman Houghton and members of the Subcommittee, it is a pleasure to appear before you today to discuss the important issue of budget scoring for tax proposals.  Under current practice, the Joint Committee on Taxation (JCT) – the provider of revenue estimates to the U.S. Congress for tax legislation either enacted or under consideration –  and the Department of the Treasury – which has a comparable role for the Administration – takes into consideration a wide variety of behavioral microeconomic responses to the incentives resulting from tax policies.  The JCT examines the effect on realizations of a change in capital gains tax rates, or the shift in consumption of gasoline in response to gasoline excise taxes. 

The purpose of my remarks today is to discuss the notion of expanding the scope of the revenue estimating process to include in some way the effect of tax policy on the macroeconomy itself – sometimes called  “dynamic scoring” – including any such macroeconomic effects on receipts.  Under a dynamic scoring approach, revenue estimates would explicitly incorporate not just individual or firm-level responses to tax-based incentives, but also changes in the overall scale of economic activity as a result of the tax policy.  That is, revenue estimates for tax changes might incorporate current and future changes in the level of Gross Domestic Product and tax bases such as aggregate earned income, aggregate corporate profits, dividends, and so forth.

Five observations frame the debate over dynamic scoring.  First, the idea of dynamic scoring is conceptually correct.  The basic notion in revenue estimation is to calculate the yearly revenue – from all relevant sources – over the appropriate budget window under current law.  To do so requires evaluating the economic activity that would prevail using current tax rules, redoing the calculation using the tax code as modified by the proposal (which clearly requires knowing the economic activity – including all relevant tax bases – under the alternative tax rules), and comparing – on a year-by-year or other basis – the revenue in the latter to the revenue in the former.  In doing so, changes in revenues from all sources would enter the revenue estimate without constraints such as a fixed macroeconomic baseline.  So, for example, if one were to switch to a broad-based consumption tax, some economists estimate that the capital stock would rise by 14 percent over the first eight years, with GDP rising by 4 percent.  The increase in wage, dividend, interest, and other sources of income embodied in these macroeconomic changes would be one source of additional revenue.  Of course, in practice estimating these steps is fraught with difficulty.  Still, these operational challenges, to which I will return below, should not disguise the basic objectives.

The second observation is that dynamic scoring represents additional information about the tax policy process.  As you know, a cost of the tax system is the distortion that taxes cause to incentives to undertake a wide range of economic activities – work, saving, investment, and so forth.  The distortion causes GDP to be lower than it would be in the absence of the tax system, or at least lower than it would be in the presence of a more efficient tax system.  Accordingly, a dynamic scoring process would reflect the reduction in deadweight loss (the economic activity foregone due to tax distortions) and increase in GDP as one part of the revenue consequences of the tax policy.  For this reason, adding this information aids policymakers in making the right choices for the economy, and policy decisions should reflect economic effects as well as revenue effects.

More mechanically, it is straightforward to conduct a revenue estimate using existing methods and supplement these estimates with an “impact statement” that shows the macroeconomic consequences and the possible related revenue effects.

The third observation is that dynamic scoring does not make sense for every tax proposal. For certain tax policy changes – substantial reductions in marginal tax rates, broad-based investment incentives, etc. – there are likely to be shifts in aggregate labor supply, saving, entrepreneurial ventures, composition of compensation, investment, and so forth substantial enough to alter both the path of the economy, and the level and time path of receipts.  As an economist, I think of this in a benefit-cost framework.  Dynamic scoring is harder, and thus more “costly” in some general sense.  For this reason it should be restricted to those circumstances in which it has real costs for the macroeconomy.  We should examine the impact on the macroeconomy in those circumstances in which conventional scoring rules can reasonably be expected to give a misleading picture of both the overall revenue effects over the relevant budget window, and the growth or transition of revenues on a year-by-year basis. 

The fourth observation builds on the previous two: Because not every proposal merits full-blown dynamic scoring, and because the macroeconomic consequences can be viewed as a supplement to (as opposed to a substitute for) current procedures, there is no need to embed dynamic scoring in the existing budget process.  Instead, for those proposals that meet the two criteria discussed earlier, the conventional scoring can be supplemented with an impact statement.  This will be useful in two ways: Policies that merit an impact statement will stand out from other tax changes, and the impact statement will be useful in guiding priorities and decision-making.  Accordingly, inclusion of an impact statement will likely have a real effect on the policy process.

The final observation is that dynamic scoring of tax proposals is difficult.  The empirical tax and economic modeling capability necessary is quite demanding.  To get a flavor of the challenge, consider dynamic scoring of a proposal to eliminate the double-taxation of dividend income.  Specifically, suppose that the policy were to be enacted this year, but not become effective for two years.  And, suppose further that full implementation of the proposal was phased in over a period of several years. 

A model suitable for dynamic scoring would necessarily need to permit the announcement of the policy to affect corporate financial policy and investment, household saving and portfolio decisions, and the resulting macroeconomic consequences for interest rates, equity prices, saving, investment, and GDP.  Further, the model necessary would distinguish between the ultimate effect when the policy had been fully phased in, and the transition path as the policy is incrementally implemented.  Economic projections informing the revenue consequences on a year-by-year basis should reflect households’ and businesses’ judgments regarding the timing of their activities in response to not only the tax incentives, but also the economic environment – which, of course, is in part influenced by their decisions.  Obviously, this is a difficult task.  (Of course, the ease of a caclulation does not make it correct; the difficulty of dynamic scoring is a not, per se, an indictment of its desirability.)

Given the inherent difficulties, one can anticipate that different modeling strategies will yield alternative estimates.  Some view this as an insurmountable impediment to the entire notion.  In contrast, I see no reason why multiple impact statements might not be produced using a variety of modeling techniques for any single tax proposal. 

Revisiting the conventions used for evaluating tax proposals is a valuable exercise.  I thank the committee for holding this hearing.  More generally, I think it is important to recognize the history of efforts in this area by this committee, the Committee on Ways and Means, the Joint Committee on Taxation through its 1997 Symposium on Tax Modeling, and many others.  The Administration looks forward to working with Congress on this issue.

Thank you, Mr. Chairman, and I look forward to discussing with you this important topic.