The 1993 tax changes are the latest in a series of important revisions to the U.S. tax code. Throughout this process, the debate has included questions about the appropriate level and structure of capital income taxation. Despite considerable research, the effects of capital income taxation are still unclear and sometimes difficult to understand. During the past few years, I have continued my research on the subject, both alone and with several coauthors.
Capital Gains Taxation
Economists have paid considerable attention to the capital gains tax in recent years, because it falls mainly on taxpayers with higher income, has a strong impact on individual behavior, and is complex in design. My research attempts to measure empirically the effects of capital gains taxation on taxpayer behavior, and to consider the efficiency effects of alternative schemes for taxing capital gains.
Over the years, taxpayers appear to have responded strongly to changes in the capital gains tax rate. For example, when a tax rate increase was scheduled to take place in 1987, realizations of capital gains roughly doubled in 1986. A more subtle and controversial question, though, is whether tax-induced increases in realizations arise from shifts in timing, rather than from long-run increases in the rate at which gains are realized. The distinction is important for determining the long-run revenue effects of changing capital gains tax rates.
In two empirical studies based on U.S. time-series data,(1) I estimate that a considerable portion of the response can be attributed to anticipated changes in the capital gains tax rate, rather than to the level of the tax rate. These results suggest that a considerable share of any revenue gain generated by a capital gains tax reduction will be short-lived.
Most of the distortions and complexities associated with capital gains taxes, such as the "lock-in" effect that discourages realizations, are attributable to the fact that capital gains are taxed on realization, rather than on accrual. Yet, moving to the taxation of gains on accrual has been hindered by two factors. First, the gains themselves often cannot be determined accurately until they have been realized. Second, taxpayers may lack the liquidity to pay taxes on appreciating assets before such assets are actually sold. Building on a much earlier contribution by William S. Vickrey,(2) I show that there is a scheme that taxes assets only when they are sold, but simulates the effects of a system of accrual taxation by effectively charging interest on deferred taxes. This system of "retrospective" taxation works even if the government cannot identify the pattern of actual accrual of gains.(3)
Would taxing capital gains on accrual increase economic efficiency? The issue is complicated, because lessening one distortion could worsen another. Capital gains taxes don't simply distort the timing of realizations; they also discourage saving. It is possible, for example, that the lock-in effect induced by taxing realizations forces additional saving, as investors discouraged from balancing their existing portfolios are driven to provide balance through additional saving. Hence, alleviating the lock-in effect also might reduce saving.
To consider this and other propositions regarding capital gains tax reform, I developed a numerical simulation model involving both saving and portfolio choice decisions.(4) I found that a move to accrual taxation does reduce national saving, but still increases economic efficiency. The same analysis suggests that a simple reduction in the capital gains tax rate also would reduce national saving, even if capital gains realizations responded strongly to the rate of tax.
Taxation and the Cost of Capital
Capital income taxes influence the level and composition of investment through the required beforetax rate of return, or the cost of capital. In my research, I attempt both to measure the cost of capital and to consider how different tax provisions influence it. …