Contributions are limited to $2,000 yearly, but eligibility to contribute is phased out for single taxpayers with adjusted gross incomes between $95,000 and $110,000 and for joint filers with adjusted gross incomes between $150,000 and $160,000. The law also allows a holder of a conventional IRA to pay taxes on it and convert it to a Roth IRA.

Both a taxpayer and spouse can contribute to Roth IRAs if their income is at least equal to the contributions. People can contribute to Roth IRAs even if they participate in a pension plan or contribute to conventional IRAs.

Proponents of Roth IRAs have long contended that the additional economic growth stimulated by funds from Roth IRA investments will generate new tax revenue that will more than offset any loss in taxes from making withdrawals tax-free. ³⁸

The other major revenue raiser was an increase in the cigarette tax from 24 cents per pack to 39 cents. Though not originally part of the budget agreement, the provision was added following announcement of the tobacco industry's decision to settle myriad lawsuits by paying $368 billion. The settlement greatly eroded the tobacco industry's ability to forestall cigarette tax increases through lobbying and political contributions. With smoking and the tobacco industry under political assault, a cigarette tax increase was simply an expedient way to raise revenue and offset tax cuts.

It is ironic, given the heated debate over the distribution of the tax package's benefits to people at various income levels, that the highly regressive increase in the cigarette tax passed almost without debate. A study by KPMG Peat Marwick found that families making less than $30,000 per year pay more than half of all cigarette taxes - a tax burden almost five times more as a percentage of income than that on families earning over $60,000 [see Figure III]. According to a more recent Tax Foundation analysis, those with AGIs under $15,000 will pay 34 percent of the tax increase and those with incomes under $30,000 will pay 60 percent. ⁴⁰ This is consistent with research showing that the incidence of smoking rises as incomes fall. ⁴¹

The ease with which Congress was able to increase the cigarette tax by more than 60 percent, almost as an afterthought, suggests that larger increases will come. There are a number of proposals to increase cigarette taxes to recoup the costs smokers impose on society, such as increased expenditures to treat smoking-related diseases. But the truth is that smokers already pay their own way. After adjusting for inflation, the direct cost of smoking - 21 cents per pack in 1995 dollars - falls well below the 1995 average total of 63 cents per pack in federal and state taxes. Since smokers overpay their incremental costs, an additional tax is unfair. ⁴²

Higher cigarette taxes are strongly promoted on health grounds, but the morality of using large increases in "sin taxes" to punish legal but socially opprobrious behavior is questionable. ⁴³

Congress responded with analyses from the JCT showing a somewhat more balanced distribution of benefits. For example, a June 17 analysis of the Senate Finance Committee bill estimated that only 43.3 percent of the tax benefits would go to the top income group. ⁴⁴

But the debate over how much of the tax cut would go to the wealthy versus the poor was really about apples versus oranges, the result of methodological differences between the Treasury and JCT going back many years. [See the Sidebar on Calculating Distributional Effects.] Because Congress and the administration were under the political control of opposing parties, it was often implied (incorrectly) that the conflicting analyses simply represented different political biases.

Figure IV shows how the two agencies scored the impact on federal revenues from the Taxpayer Relief Act. The JCT forecast a substantially larger revenue loss from the legislation than did the Treasury. The difference between the two estimates derived largely from the different estimates by Treasury and the JCT of the effects of the capital gains provision. Figure V indicates how the JCT and Treasury scored the revenue effect of the final capital gains provision. There was a difference of almost $25 billion between the two estimates for the period 1997-2007. [Also see Appendix Tables I and II.]

These sharply contrasting revenue estimates had important effects on the distributional analyses. The reason is that both Treasury and the JCT calculate distribution on a static basis. In other words, they assume that assets that are sold solely due to the lower capital gains rate would have been sold even in the absence of any tax change. Thus the larger the unlocking effect of cutting the capital gains tax, the bigger the tax cut for those with capital gains as reflected in the distributional tables. Because Treasury estimates a much larger unlocking effect than the JCT, it allocates a much larger tax cut to the wealthy in its distributional tables than does the JCT.

As Figure VI and Appendix Tables III and IV indicate, a result of these methodological differences is that the estimated revenue loss from the Taxpayer Relief Act is 50 percent higher in Treasury's distribution table than in the JCT's. Indeed, Treasury's estimate of the tax cut distribution by income group is almost twice as much as the projected revenue reduction in its own revenue estimate, shown in Figure IV. By contrast, the distributed tax cut in the JCT's table is much closer to its estimate of the revenue loss.

While there is no evidence that the Treasury or the JCT manipulate their distribution tables for political purposes, the ultimate result of their different methodologies was to bolster the political agendas of their respective masters. The JCT's methodology showed a smaller tax cut for the rich, thereby aiding the goals of the Republican Congress. Treasury's analysis supported the view of the Clinton administration that the tax cut was tilted toward the rich, even though Treasury's own revenue estimate scored the capital gains provision as a net revenue raiser over the entire forecast period.

In recent years, academic researchers have begun to pay greater attention to the problems inherent in calculating the distributional effects of tax changes over different income levels. ⁴⁵ Debate over the 1997 tax bill clearly shows that more work needs to be done.