The choice of a base year has only a small effect on estimated effective tax rates. The results discussed in the body of this paper derive from incomes in 2001. Incomes in that year were generally lower than those in 2000, particularly at the upper end of the distribution. Nonetheless, substituting incomes in 2000 for those in 2001 and repeating the analysis yields only slightly higher estimates of effective tax rates, with much of the observed difference reflecting a change in the amount of capital gains realized.

Economic circumstances differed markedly between 2000 and 2001 (see Table A-1). Average income for households rose sharply during the 1990s--a period of rapid economic growth and a booming stock market--peaking at $76,200 in 2000. In 2001, the economy went into recession, the stock market fell, and average income dropped 6 percent, to $71,800. The decline in income varied across the income distribution. While average income for the top quintile fell by 10.5 percent, from $202,000 in 2000 to $182,700 the following year, it fell by about 1 percent for the fourth quintile and by less than 1 percent for each of the three lower quintiles.

A large part of the difference in incomes between 2000 and 2001 resulted from capital gains realizations, which fell from $644 billion, or 6.6 percent of gross domestic product (GDP), in 2000 to $344 billion, or 3.5 percent of GDP, in 2001. That $300 billion drop constituted about two-fifths of the total decline in income between the two years. At 3.5 percent, realizations for 2001 were close to the historical average of capital gains relative to GDP.

Shifting the base year for incomes from 2001 to 2000 moves estimated changes in effective tax rates by no more than one-tenth of a percentage point for any quintile in 2004 (see Table A-1). The differences are greater for households with the highest incomes: the top 1 percent of households shows a 6.8 percentage-point reduction using 2001 incomes but only a 6.1 percentage-point drop using 2000 incomes. That difference results from the fact that some of the tax reductions for 2004, such as the expensing provisions, are measured in fixed dollar amounts and are not a function of income. Those fixed dollar amounts cause a bigger percentage-point change in effective tax rates when based on the lower incomes in 2001.

In 2006, the effects of using incomes from alternative years are greater, primarily for the top quintile. Effective tax rates again differ by no more than one-tenth of a percentage point for the first four quintiles, but the top quintile shows a reduction of 1.4 percentage points using 2001 incomes, compared with a reduction of 1.7 percentage points using 2000 incomes. The difference rises to 0.5 percentage points for the top 1 percent of households. Those differences derive from different levels of capital gains realizations between the two years interacting with the Jobs and Growth Tax Relief Reconciliation Act's reduction in the maximum tax rate on long-term capital gains. Adjusting the levels of realizations in the 2001 data to be consistent with the higher 2000 levels eliminates most of the difference in effective tax rates (see the bottom panel of Table A-1). Thus, the larger impact of legislative changes in 2006 measured using 2000 incomes results from the unusually high level of capital gains realizations in 2000.

In 2010, the results are almost identical using the 2000 data and the 2001 data, within one-tenth of a percentage point for all groups. The rate reductions on capital gains are not in effect in that year, and the impact of tax changes measured in fixed dollars is very small. By 2010, then, the choice of a base year for incomes has little effect on the changes in effective tax rates.

The choice of a base year has a greater impact on the changes in the share of taxes paid by each income segment (as opposed to the effective tax rate for each segment), although the impact is still not dramatic (see Table A-2). The largest differences are for the top quintile and the subcategories within it. For example, under 2004 law, the share of taxes paid by the highest quintile falls by 0.6 percentage points using the 2001 data compared with 0.4 percentage points with the 2000 data.

Some of the difference in shares of taxes resulting from the choice of a base year comes from the difference in realized capital gains. Under 2006 law, the 2001 data show a 0.1 percentage point increase in the share of taxes paid by the top 5 percent of households, while the 2000 data show a decrease of 0.3 percentage points. However, if the 2001 data are adjusted to reflect the higher realizations in 2000, the changes in shares for the top 10 percent, 5 percent, and 1 percent of households are identical.

Table A-1.

Table A-2.