May 2006

Tax Reform and Dynamic Analysis

One of the more hotly debated policy questions surrounding tax reform is how revenues are affected by changes in the tax code. The current standard for assessing such changes does not take into account growth effects from tax cuts, which in turn boost revenues. Ignoring such effects can make tax cuts appear more costly to government coffers than they ultimately are. To provide a clearer picture of the tax reform alternatives detailed in the final report of the President’s Advisory Panel on Federal Tax Reform, the Treasury Department performed an assessment--a dynamic analysis--on the panel’s reform alternatives that would incorporate possible economic growth resulting from the tax cuts.

Robert Carroll
U.S. Treasury

The new dynamic analysis division is critical to evaluating whether tax reform proposals meet the president’s objectives of being simple, fair, and pro-growth. Dynamic analysis focuses on the economic benefits of tax policy changes; it is distinct from dynamic scoring. The analysis focused on the long-run effects of the simplified income tax, the growth and investment tax, and the progressive consumption tax, using multiple models to reflect the sensitivity of the results to underlying assumptions. All reforms were found to increase output in the long run, with the largest gains registered where taxes on capital income were reduced the most.

Craig Johnson
U.S. Treasury

Three standard models were used to evaluate each tax proposal. All assumed perfect competition, no inflation or change in the money supply, and standard Cobb-Douglas production functions. The first was a Solow model with one sector. The second was a single-sector infinite horizon Ramsey model with household-level optimization of saving and labor supply, constant elasticity utility function, perfect foresight, and infinitely elastic long run savings. The third was an overlapping generation (OLG) model with household life cycles of fifty-five years, perfect foresight, four different sectors, and moderate international capital flows. The Solow model, with its fixed labor supply and moderate savings response, represented the low end of the economic impact of tax policy changes, while the Ramsey model, with moderate labor response and infinitely elastic savings, represented the high end.

Each policy simulation began with a baseline steady state based on current policy. Revenue neutrality, including transition relief to owners of capital, was maintained each year by adjusting average and marginal tax rates. The three reforms under consideration were the progressive consumption tax (PCT), essentially a subtraction-method value-added tax with progressive wage tax rate structure; the growth and investment tax (GIT), similar to the PCT but with a tax on dividends and capital gains; and the simplified income tax (SIT), which uses income as its base but lowers rates on corporations and individuals. The latter two proposals also include tax-free savings accounts.

The PCT led to the highest gains in outcome (measured by net national product), followed by the GIT and SIT, respectively. The growth potential of each plan is related to how much it lowers the tax on capital income, and the PCT does this the most. Each proposal yielded the highest benefits under the Ramsey model and the lowest under the Solow model. Under the PCT and GIT, the labor supply was influenced by income effects arising from the effective tax on old capital and the resulting decreases in wealth, but the inclusion of transition relief moderated these labor supply increases. The SIT taxes both old and new capital, but evens out tax burdens across production sectors and thus shifts capital to the corporate sector. This increases efficiency and broadens the base, allowing labor taxes to fall and labor supply to increase. However, the SIT reduces taxes on capital income, the least of the three options, and therefore causes the smallest growth in the capital stock. 

These models are limited in that they do not completely take into account international capital flows. Also, the Ramsey and OLG models are frictionless, meaning that it is easy to change investment and shift labor over time. Accordingly, short-run results might be overstated. Additionally, the models do not account for risk, and so might overstate savings responses. On the other hand, they also fail to include certain benefits, such as reduced compliance costs, more efficient allocation across sectors (in the Ramsey and Solow models), and gains from fixing the biases in favor of debt financing and non-corporate capital.

Alex M. Brill
House Committee on Ways and Means

A range of organizations and governmental bodies have looked at the budget impacts of large scale tax reforms. House rules require this analysis of any large tax policy coming out of the Ways and Means Committee. This paper contributes a great deal to the goal of transparency in evaluating tax policies as well as to the overall refinement and improvement of the tax code and tax models. The OLG model is the most complex and probably the most appropriate of the three employed.

Unsurprisingly, the effects of fundamental tax reform were found to be much larger than those from incremental reforms. The benefits may not even be fully captured in the OLG model because it does not consider compliance costs and may underestimate the elasticity of labor supply. However, the results may also be overestimated in terms of the likelihood of comprehensive reform actually being enacted. Revealingly, the tax panel itself was politically constrained in what policy choices they could examine--for instance, they were not allowed to consider eliminating the tax subsidies for housing and charitable giving.

Rudolph G. Penner
Urban Institute

The paper demonstrates the potential gains of tax reform in a realistic way. Especially promising is the possibility of leveling the playing field between various types of capital. The models used here do not reflect a lot of the gains from dealing with the distortions in taxation of housing and the health insurance exclusion. Also, the paper might underestimate the effects of tax reform on saving--a consumption tax could lower the income of high spenders while enhancing the income of high savers. Reliance on Cobb-Douglas production functions also may underestimate the importance of capital to economic growth.

However, ignoring foreign capital inflows in the Solow and Ramsey models is problematic in that increases in capital investment might simply be replacing these flows rather than generating new economic activity. The economic impact of tax reform might also be impacted by expectations regarding the federal budget, and some revenue estimates would have been helpful. In the end, the issue is whether voters would exchange immediate benefits such as the mortgage interest deduction for the promise of faster growth over the long term, which is doubtful.

Kevin A. Hassett
AEI

The U.S. Treasury’s efforts reinforce the commonalities among the tax policy community. The models in the paper are highly stylized compared to other, more realistic models used in other work. These models, and those used by the public finance community in general, are also quite old and far removed from the latest developments in the literature, failing, for example, to consider imperfect competition or uncertainty. It would be useful to look at past tax policy changes and calculate whether dynamic analysis would have improved forecasting in order to convince people of the need to perform dynamic scoring in the future. A good use of the results in the paper would be to show Congress the importance for economic growth of getting tax reform right. A bad consequence might be that dynamic scoring could convince Congress that they have more money to spend, and thus mitigate any revenue gains identified in the scoring process.

AEI intern Matt Perlman prepared this summary.