Innovation and the long-run elasticity of total taxable income.

• Author: Russo, Benjamin

1. Introduction

   The elasticity of taxable income with respect to the income tax determines revenue responses to changes in the tax rate and influences the welfare effects of the tax. As a result, measurement of this elasticity is a continuing focus of research on tax policy. (1) Most empirical estimates of taxable income elasticities are based on short-run analyses. However, long-run fiscal policy effects may differ from short-run effects in sign as well as in size. (2) The sign reversals arise from long-run general equilibrium relationships. The empirical literature has often cited the desirability for long-run measurement. (3) However, in this regard nearly no work has been reported. This paper derives long-run general equilibrium elasticities of taxable income with respect to income taxes and discusses policy implications.

   Before the advent of the concept "taxable income elasticity" economists often estimated labor supply and saving elasticities in an attempt to quantify the behavioral and efficiency effects of taxes. Many empirical estimates indicate labor supply and saving responses to aftertax wage rates and interest rates may be relatively small. (4) However, Lindsey (1987) argues that the important behavioral responses to tax rates include more than just changes in labor supply. For example, a change in the tax rate may lead to changes in capital gains realizations and itemized deductions. Feldstein (1995) argues that the behavioral responses to taxes also include changes in portfolio investment in the short run and changes in the location of work and types of jobs in the long run. (5) These authors suggest that changes in taxable income reflect behavioral responses to taxes more reliably than does labor supply, and they develop the concept of the elasticity of taxable income to quantify such changes.

   The literature on empirical estimation of taxable income elasticities has generated many important insights into behavioral responses to income taxes. Nevertheless, factors exist that could cause empirical estimates to overstate or understate true tax responsiveness. For example, some tax avoidance schemes can cause shifts in the timing of economic decisions without affecting real economic activity (Slemrod 1996). In this case, empirical estimates of taxable income elasticities could overstate tax responsiveness. In addition, if long-run general equilibrium relationships are not captured by empirical data, empirical estimates could understate tax responsiveness. To see this, note that in general equilibrium initial tax responses in one market can be reinforced by responses elsewhere. Suppose, for example, that a change in tax rates increases labor supply and saving, initially, by small amounts. Over time, larger saving increases the capital stock and, therefore, the marginal product of labor. Higher wages, in turn, could stimulate labor supply. Transactions and adjustment costs, as well as uncertainty, could cause these inter-relationships to take a long time to develop, so it may be difficult to capture them in standard empirical analyses and in event studies in particular.

   In this paper, I attempt to avoid some of the difficult issues raised by Slemrod (1996) and, at the same time, capture long-run general equilibrium relationships that seem likely to affect tax responsiveness. To these ends, I use a Computable General Equilibrium (CGE) simulation model to construct long-run taxable income elasticities.

   A mainstay of previous CGE tax analysis is a model in which physical capital investment is motivated by profits, but in which innovation is exogenous. (6) However, innovative activity must vary with monopoly profits, to some extent, because profits are required to repay the large upfront costs of Research and Development (R&D). And it seems reasonable to assume that some, if not many, entrepreneurs expect to earn economic profits from technical knowledge their firms create. (7) Technical knowledge is at least partly non-rival, so the marginal cost of providing an innovative good to an additional user tends to be small. Therefore, the extent of the market for innovative goods can have a large effect on after-tax profits and the incentive to innovate. This indicates that the economy could be more tax responsive if innovation is endogenous rather than exogenous. In addition, saving and innovation share a complementary relationship under endogenous innovation: An initial tax-induced increase in innovation tends to encourage more saving, which in turn could reinforce the initial increase in innovation (Lin and Russo 2002). This possibility also indicates that the economy would tend to be more tax responsive under endogenous, rather than exogenous, innovation. This paper models endogenous innovation and then compares responsiveness under the two regimes.

   While empirical analysis has some weaknesses in terms of estimating taxable income elasticities, I am mindful that CGE models also have important limitations. They are highly stylized, and their solutions require strong assumptions. CGEs are calibrated to mimic actual economies by culling parameter values from the empirical literature, so they inherit some of the problems from empirical estimates. No single technique appears capable of producing complete answers to questions about the effects of tax policy. It is important to note, also, that I study only steady states. Behavior along transition paths can produce important insights, especially when relationships are highly non-linear, as they are in the model used here. The analysis of transitions under endogenous innovation is difficult and is left for future work.

   With these qualifications in mind, the simulation results reported below are suggestive of the following points regarding taxable income elasticities:

   (i) Long-run elasticities can be relatively large, even if initial impacts of taxes on labor supply and saving are small;

   (ii) Long-run elasticities with respect to the corporate tax can exceed uncompensated elasticities with respect to the individual tax by very large margins; and

   (iii) Elasticities are much larger under endogenous innovation than under exogenous innovation.

   The second point indicates that reducing the corporate income tax by a percentage point could increase the efficiency of the tax system more than reducing the individual income tax by a percentage point, even if the initial tax rates are the same. If true, policy makers concerned with improving the efficiency of the tax system may obtain better results by reducing the corporate income tax rate than by adjusting the individual income tax rate. The third point indicates that elasticities tend to grow as value added in the innovative sector grows as a share of national income. Relatively large elasticities with respect to the corporate tax, together with the complementary relationship between saving and innovation, could lead to substantial shifting of tax incidence away from capital in the long run. (8) If this occurs, capital income could become harder to tax, and progressive tax structures may become more difficult to sustain as economies evolve toward higher levels of innovation.

   The simulations indicate at least two additional results worth mentioning. First, the R&D tax credit could magnify taxable income elasticities. Second, a form of tax avoidance with no apparent real short-run effects, namely, a tax-induced shift from corporate to non-corporate structure, can produce real long-run effects. In particular, a change in the structure of taxes that encourages producers to shift from corporate to non-corporate form may reduce the elasticity with respect to the corporate tax and increase the elasticity with respect to the individual tax.

   The next section reviews literature that is relevant to the issues I study and the techniques I use. Slemrod (1998) argues the elasticity of total taxable income should be the central concern in studies of tax elasticity. This point seems particularly relevant here because in general, equilibrium behavioral responses to individual taxes can affect corporate income, and behavioral responses to corporate taxes can affect individual income. (9) Therefore, I focus on the elasticity of total (individual plus corporate) taxable income. Section 3 describes the model used in the simulations. Section 4 describes the results of benchmark numerical simulations of elasticities of total taxable income with respect to the individual income tax and, separately, with respect to the corporate income tax. Compensated and uncompensated elasticities are reported for the individual income tax. I also compare responsiveness under endogenous and exogenous innovation and report on an extensive sensitivity analysis. Section 5 summarizes the paper. Appendix A defines all symbols used in the text.

2. Literature

   The original papers in the taxable income elasticity literature argue that income can be responsive to taxes even if saving and labor supply are relatively unresponsive (Lindsey 1987; Feldstein 1995). Although early precursors of the elasticity literature did not study taxable income per se, their work has a similar implication.

   Harberger (1962) uses a general equilibrium model with fixed capital stock to show how the incidence of the corporate income tax may fall on capital, on labor, or on both. The final resting place of the tax depends on relative capital intensities and capital/labor substitution elasticities in the corporate and non-corporate sectors. This is a crucial economic possibility per se. However, the implication relevant to the current paper is that if the tax causes capital to migrate to a relatively unproductive sector, the corporate income tax could tend to reduce taxable income.

   Feldstein (1974) extends Harberger's analysis to a long-run setting, with exogenous saving and labor supply. The incidence of a tax on capital income may not rest fully on capital, even if the personal...