Making Tax Policy: A Variance Ratio Approach to Measuring Tax Incidence
Whicker, Marcia L., de Lancer Julnes, Patria, Williams, Daniel W., Journal of Public Budgeting, Accounting & Financial Management
ABSTRACT. This paper briefly examines previous Lorenz Curve approaches to measuring tax incidence, including the Gini coefficient approach, and Suits's S measurement of tax effects. Both approaches have advantages, but are found lacking in providing useful simple information to policy makers and citizens, especially when comparing proposed but not yet implemented taxes or tax changes. A variance ratio measure, TIC, is suggested to evaluate proposed tax policy changes. This measure uses income data and tax rates to compare the variance for relative tax shares to the variance for income shares. In several examples, TIC performs as expected.
The two chief issues in budgeting are expenditures and revenues. In recent years, revenue discussions have focused on tax reductions. The redistributive impact of proposed tax cuts has often proved secondary in both analysis and decision-making to the implementation and magnitude of tax cuts. Yet, distribution is the core of politics, identified as such by Aristotle as well as subsequent scholars. As a concept, distribution has been central to concerns of political science, economics, and sociology (Allison 1978; Rivlin 1975; Whicker and Strickland 1990; Whicker, Strickland and Olshfski 1993; Slemrod 1994). Little consensus exists on the preferred distribution of income held by individuals, although some theorists argue for equality (Rawls 1971). Increased concern about the distribution of income in the United States grew in the 1980s and 1990s as evidence mounted that distribution was becoming more unequal (Edsall 1984; Phillips 1990, 1993; Dazinger and Gottschalk 1993).
By the end of the 1980s, it was very apparent that the rich were getting richer and the poor were becoming relatively poorer, while the middle class shrunk. According to the U.S. Bureau of Statistics, the share of total income going to the top 20% of all families in 1985 was 43.5%, the highest level since 1947 when such statistics were first compiled. The bottom three-fifths of the population received the smallest share of income ever -- 32.4 %. Thus the top fifth received 11 % more than the bottom three-fifths combined. Nor is this income distribution skew racially neutral, since racial minorities continue to lag behind whites in earnings and have higher incidence of poverty.
Phillips (1990) among others has argued that changes in tax rates and shifting tax burden, in addition to differential income growth across income cohorts, has contributed to the growing disparity in post-tax income distribution. Musgrave and Musgrave (1976), and Okun (1975) have produced the classic tax incidence studies for the United States. These authors derived different estimates of incidence, depending on the assumptions they employed about shifting of tax burden from one income group to another. Under competitive assumptions, the corporate income tax, property tax, and other taxes cannot be passed forward to consumers. With competitive assumptions, the pre-1980s tax system was modestly progressive at the very top and bottom of the income distribution, but proportional over the middle income range. Changing to non-competitive assumptions so that the property tax, corporate income tax, and social security payroll tax could be passed forward to consumers causes even this small amount of progressivity to disappear (Browning, 1978).
Plainly, decision-makers need information about the distributional affects of proposed tax legislation. The purpose of this paper is to examine the information available to policy makers when enacting revisions in the U.S. tax code. It will discuss the need for a simple measure of tax structure incidence that can be readily and easily communicated to the public before tax changes are enacted. One measure -- a variance ratio approach will be suggested. Examples of how the Tax Incidence Coefficient (TIC) could be applied will be provided.
PREVIOUS EMPIRICAL ESTIMATION APPROACHES TO MEASURING TAX INCIDENCE
Various approaches to date have been used to measure tax incidence. …