Pechman's Tax Incidence Study: A Response
回应Browning对作者1985年税收归宿研究数据问题的批评，指出其推断方法不当，并提供了基于修正权重的1980年税收负担计算，展示了不同收入十分位组的有效税率。
I am indebted to Edgar Browning for calling attention to some peculiarities of the data underlying the estimates in my 1985 study. He is quite right that the ratios of transfer payments to income in the files for 1975 and later years are inconsistent with the corresponding ratios in the 1966 and 1970 files and in the Consumer Population Surveys, particularly in the lower part of the income distribution. I do not believe, however, that the baby should be thrown out with the bathwater, as Browning seems to suggest. I made available to him detailed data on the tax burdens of labor income, capital income, and consumption by income classes which provided a basis for revising my calculations for 1975 and later years. Instead, he chose to draw inferences about the progressivity of U.S. taxes since 1966 on the basis of changes in the relative importance of the various taxes, which is not appropriate for this purpose. These inferences would be correct only if there were no changes in the structure of each tax and in the composition of income in the various income classes during the period studied (for example, if there were no changes in the progressivity of the individual income tax or in the distribution of dividends by income classes).' It is straightforward to improve on this procedure. Where in the income distribution the share of transfer payments in total income is too low in 1975 and later years, the shares of labor and capital income are correspondingly high, and vice versa. To arrive at more realistic estimates of the shares for labor, capital, and transfer income, I first applied the 1970 shares to adjusted family income in each income decile in 1975, 1980, and 1985, and then made proportional adjustments in the columns and the rows alternately until they added to the correct totals.2 The incidence calculations for 1980 based on the revised weights are given in table 1 of my study, which shows the effective rates of the major taxes by deciles.3 These calculations are based on the most progressive set of assumptions (variant lc) presented in my study. Variant Ic assumes that the corporation income and property taxes are borne by capital in general, the payroll tax by labor, consumption taxes by consumers, and the individual income tax by those who pay it. I believe that this variant more nearly reflects the present state of incidence theory than any of the other variants.4