The Key Behavioral Elements Needed in a Model for Budget Estimation

Report Budget and Spending

The Key Behavioral Elements Needed in a Model for Budget Estimation

May 30, 2002 12 min read

Authors: Aldona Robbins, Gary Robbins and William Beach

Members of Congress and key leaders of the Bush Administration find themselves wrestling with one of the more important, though obscure, issues currently in public debate: how to change the process of estimating the revenue effects of budget policy changes in a way that will incorporate the effects of those changes on the U.S. economy. Though this issue seems to be narrowly defined, it is playing a central role in the larger debate over which tax policy changes would best promote economic growth and widespread prosperity.

While most economists would address this issue by estimating the effects of policy changes using a statistical model of the U.S. economy, the current practice of Washington-based budget estimation is to assume that taxes have no effect on the macroeconomy.1 In other words, the federal government's principal tax analysis groups do not use a model of the U.S. economy in creating their estimates of revenue change.

This practice, however, runs counter to real-world experience, in which the level of taxes is critical in determining how businesses organize and how people allocate the resources available. Instead of assuming these effects away, the structure of any model used in budget analysis must, at the very least, allow taxes to enter the relevant economic decisions.

Some still disagree with the whole idea of evaluating tax policy changes in terms of their macroeconomic effects, but most parties to this "dynamic scoring" debate are actually engaged in arguments over which assumptions to make when using a model of the U.S. economy. The assumptions of any model are key to the results it will produce. For example, if a model assumes that employees will not change their hours worked whether take-home pay goes up or down, one should not be surprised that the model
will predict no economic effects from increasing payroll tax rates. Similarly, if a model assumes that taxes play no role in how much capital is made available by investors, it will conclude that changing tax rates will have no effect on economic activity.

The economic properties of macroeconomic models are the crucial elements in creating an accurate forecast of how tax policy changes affect economic activity. Getting these properties right matters more to the information the model produces than how many economic concepts or years the model can report. In other words, if the economics are not right, the model will produce output that may lead the model user to the wrong conclusions.

Policy leaders in Congress and the Administration must insist on a higher, more sophisticated, and more informative level of analysis from the tax staffs that support the government's important policy work. The Joint Committee on Taxation and the Treasury Department's Office of Tax Analysis should be required to adopt immediately the practice of routine macroeconomic modeling for all important tax policy changes.

Key Elements of an Economic Model for Budget Analysis

Policymakers are right to wonder what major building blocks a model of the U.S. economy should have. Selecting these building blocks, or key elements, is important; the ones selected will largely determine the results of a model run, or simulation, of a policy change. Thus, the users of a model's output need to know the important assumptions of the model if they are to make a sensible assessment of what it reports.

The following are examples of some of the most important economic relationships that should be incorporated into a model that is used for budget and tax analysis but that are now overlooked.

  • Labor and capital combine to produce private output.
    More labor and more capital available to business should result in more output. As the rewards to labor and ownership of capital increase, workers and investors are willing to increase their respective supplies. This is how it works in the real world, and any budgetary model should exhibit this result. How capital and labor inputs relate to output is a matter of existing technology.

    Any budgetary model should be able to demonstrate that its representation of technology can replicate the historical record and comport with observed facts.

  • Labor and capital respond to changes in marginal tax rates.
    Workers supply labor based on real after-tax wage rates. In considering whether to work additional hours, workers will evaluate how much they will keep of the extra wages they earn, after taxes and inflation. Therefore, what matters are marginal, not average, tax rates. Studies that look only at the effect of average taxes on economic decisions miss the mark both on theoretical grounds and with respect to real-world experience.

    Similarly, the supply of capital is based on the return paid to owners of capital after inflation, replacement costs, and taxes. Again, it is tax rates at the margin that matter. Furthermore, the analysis must cover all taxes, not just selected federal taxes.2

    Historically, the real after-tax return to capital has been extremely stable.3 More important, this stability has persisted despite many substantial changes in the way investment is taxed. The reason for this stability is that the stock of U.S. capital adjusts very quickly in response to changes in the real after-tax return on capital. For example, lowering the marginal tax rate on capital brings forth more investment until the stock of capital increases enough to lower the rate of return to where it was before the tax cut. These adjustments occur fairly quickly; indeed, the historical record of the past 50 years indicates that the supply of capital is very adjustable after the first 5 to 10 years following a significant change in marginal tax rates.

    The constancy of the real after-tax return to capital implies that stock adjustments occur through changes in domestic saving and the placement of capital worldwide. Experience of the past 50 years shows that a 10 percent increase in the real after-tax return to capital will produce a similar increase in domestic saving. The reverse is true for a decrease in after-tax return. An acceptable budget model would incorporate the relationships among marginal tax rates, the supply of and demand for goods and services, and the supply of and demand for the factors that produce them.

  • The United States operates in a global economy.
    Changes in the volume of domestic saving, however, are usually not enough to drive the after-tax return to capital back to its historically stable level within the typical adjustment period. The rest of the change in supply must come from changes in international capital flows. Specifically, increases (decreases) in the real after-tax return result in less (more) U.S. investment abroad.

    More generally, an increase in the rate of return to an asset due to lower U.S. taxes on capital is shared with all investments, including those not directly affected by the tax change. For example, suppose taxes are cut for U.S. citizens owning physical assets in the United States. As these individuals rearrange their portfolios to take advantage of the increased returns on the newly favored assets, the rate of return on other assets that they liquidate also will increase, making them attractive to investors who cannot directly take advantage of the tax change. In other words, a tax change that initially appears limited to the United States is in reality shared worldwide. Absent this linkage, the real after-tax return to capital would not exhibit the constancy that it has exhibited.

    Clearly, changes in the ownership mix of capital sited in the United States have implications for tax collections. Adjustments from abroad, which involve changes in factor payments to foreigners, will be incorporated into U.S. tax bases. Increases in foreign ownership, for example, will lower the portion of dividends paid to U.S. citizens. However, these changes will be small in comparison to the increase in tax bases resulting from the higher growth due to more foreign investment in the United States. A suitable budgetary model must not assume that the U.S. economy is closed. It must allow international capital flows to be affected by changes in budget policy.

  • Federal Reserve policy matters.
    An acceptable budget estimation model must assume that monetary policy significantly influences the general price level and, thereby, nominal interest rates. Simulations done with the model should assume that the Federal Reserve would undertake whatever policies are necessary to maintain prices at their baseline levels. Guessing the future course of monetary policy, as a function of alternative budget policies, is simply an added complication. An acceptable budgetary model would not assume that the Federal Reserve would misbehave. The return on financial assets--interest rates--depends on the real after-tax return to capital and on inflation. Assuming the Federal Reserve holds constant the general price level, interest rates will change only with the return on physical assets.4
  • Tax and spending policies affect the federal debt.
    Changes in tax and spending policies should affect the level of government borrowing and debt symmetrically. The requirement that interest rates be primarily tied to the real rate of return to capital means that changes in borrowing principally affect domestic portfolios, since there is no incentive for the portfolios of foreigners to change. For example, while an increase in spending or a decrease in taxes may put money in peoples' pockets, it must also take that same amount out of the pockets of lenders. The model should show little or no first-round income effect from either tax or spending changes. The major effect of such changes flows from the initial price effect and subsequent adjustment to the new prices.
  • Federal purchases of goods and services affect private output and GDP.
    Some government purchases of goods and services count as part of the nation's output. The compensation of government employees adds directly to the value of the gross domestic product (GDP). Government employment also raises the total demand for labor. Thus, expanding government employment bids up the price of labor in general. Higher labor costs in the private sector reduce private output, offsetting part of the gain attributable to the government increase.

    Government capital expenditures should show a similar effect. In this case, GDP goes up by less than the value of the expenditure because, under Department of Commerce accounting rules, government capital yields only asset depreciation. Again, the direct increases are generally offset by reductions in private output due to higher factor costs. To the extent that either of these types of government spending increases private-sector productivity, the private-sector loss is lower; but this must be the exception rather than the rule. An acceptable budget model should show that a dollar increase in government employment or investment would lead to less than a dollar increase in GDP.

  • Federal subsidy programs cause price distortions.
    Direct and indirect (through the tax system) subsidy programs artificially reduce certain prices facing consumers and businesses. The lower price shifts demand away from other goods and services toward the favored items. As a result, more of the favored goods and services are produced than would otherwise be the case, leading to a less efficient allocation of resources. The price of the subsidized good increases with demand, and the prices of other goods decrease. In markets where there are supply difficulties--the usual justification for intervention--subsidies are often counterproductive. For example, tax credits for higher education typically lead to increases in tuition that can offset most or all of the subsidy. An acceptable budget model should show the negative GDP effect of the price distortions as well as the increased spending on the subsidized item.
  • Federal transfer programs also affect price levels.
    Government transfer programs operate in a manner similar to subsidies. They generally target a group of individuals rather than particular goods or services. Increased spending by the subsidized group raises the price of goods for everyone, producing a similar GDP effect. An acceptable budget model should show the negative GDP effect of the price distortions as well as the increased spending by the subsidized groups.
  • Federal mandates and regulatory programs act like a shadow tax on the economy.
    Mandates and regulations require businesses and individuals to engage in economic transactions they otherwise would not undertake. This acts like a shadow tax on the economy. To maintain the rate of return to capital (and in some cases labor), the marginal value product of capital (or labor) must increase to comply with the government requirement. This increase in factor costs should result in reduced GDP. Externalities, such as pollution from a coal plant, are the typical justification for government mandates and regulations. The benefits from this kind of government intervention are generally not measured because they do not show up in measured GDP. An acceptable budget model should show the negative effect on GDP and the price distortions caused by regulations and mandates. Estimates of the value from reducing externalities could be shown separately as a cost-benefit analysis, much as the Office of Management and Budget does for major rule changes.


The current practice among the federal government's principal tax analysis groups is to assume that taxes have no effect on the macroeconomy. In other words, they do not use a model of the U.S. economy in creating estimates of revenue change based on tax policy changes. This assumption runs counter to real-world experience; the level of taxes is critical in determining how businesses organize and how people allocate their resources. The structure of any model used in budget analysis at the very least must include the effects of taxes in relevant economic decisions.

Finding an acceptable set of assumptions or properties on which to base an economic model for such budget analysis is not an impossible task. Economics is a well-developed social science in which a surprisingly large body of theory is considered generally noncontroversial. Macroeconomics is a developing branch of economics, but its areas of controversy should not prevent policymakers from relying on those parts that can shed light on important policy questions. Indeed, if the fear of making incorrect assumptions prevents the use of models that contain accepted economic thinking, the debate over policy change will become less informed and more confused.

Policymakers and decision-makers in Washington must insist that those who estimate the effects of a tax policy change on the federal government's revenues incorporate the economic effects of such change in their analysis. Unless such organizations as the Joint Committee on Taxation and the Office of Tax Analysis include the economic effects of policy change in their analyses, their estimates of revenue change will be more inaccurate than they otherwise would or should be.

Gary Robbins is President and Aldona Robbins is Vice President of Fiscal Associates in Arlington, Virginia. William W. Beach is Director of the Center for Data Analysis at The Heritage Foundation.

1. Government budget estimators sometimes build in second-order effects, such as the timing of capital gains realizations in response to a change in capital gains tax rates. Ignored, however, are the first-order effects on growth, investment, and employment, which can be considerably larger.

2. Taxes at the state and local levels account for one-third of all U.S. taxes.

3. Between 1954 and 1999, the real after-tax rate of return to U.S. capital averaged 3.7 percent, with a standard deviation
of 0.5 percent.

4. Conventional wisdom to the contrary, interest rates do not change in response to changes in the level of budget surpluses or deficits unless there is an effect on the real after-tax return to capital or inflation.


Aldona Robbins

FormerVisiting Fellow

Gary Robbins

Former Visiting Fellow

William Beach

Senior Associate Fellow