Library / Publications
Designing a Work-Friendly Tax System
Options and Trade-Offs
The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.
Note: This report is available in its entirety in the Portable Document Format (PDF).
Taxes influence our choice between work and leisure. High effective marginal tax rates may discourage and certainly distort our incentive to work. Marginal rates differ from average tax rates in that they quantify the rate that applies to a worker's last dollar of earnings. A tax filer earning $18,000 in 2004 and in the 10 percent tax bracket would face a marginal tax rate of 10 percent on each additional dollar earned above $18,000.1 This is true even though most of her income below $18,000 may not be taxablein other words, her average tax rate would be well below 10 percent.
Suppose further that she has two children and receives the earned income tax credita generous tax credit targeted to low-income families that decreases by 21.06 cents for every dollar earned above $14,040. This benefit phaseout rate is hidden to her in the sense that it is not associated with any tax bracket on her tax forms. However, her effective marginal tax rate includes it, amounting to the sum of the 10 percent income tax plus the 21.06 percent EITC phaseout, for a total rate of 31.06 percent. But this calculation is still not complete. We have not included the child tax credit that would lower her effective marginal tax rate (because this credit is still phasing in at her salary level) or her Social Security payroll taxes that would raise it again. We see, though, that effective marginal tax rates may differ from statutory tax rates by a good dealthey may be both lower and higherand they are confusing to assess because our panoply of tax statutes and programs place differing statutory and hidden rates in effect at nearly every income level.
It is this effective marginal tax rate that economists believe strongly influences a household's decision to engage or not engage in more work2regardless of whether the household is aware of the tax rate. The higher the tax rate, the less the household will gain from additional work and (many believe) the more likely the household will substitute leisure for work or find some way of avoiding the tax. In particular (although, not without debate), the empirical evidence tends to show that high effective marginal tax rates can discourage work by low- and moderate-income individuals, especially those who are trying to work their way off the welfare system.3 Unfortunately, the current federal tax system often imposes its highest effective marginal tax rates on just those individuals. The purpose of this paper is to suggest some options for reducing those high effective marginal tax rates.
The reform options described in this paper are in line with some past work in the field addressing high tax rates confronted by low-income families.4 One approach would replace the current earned income tax credit with a $2,000 per working parent credit and a refundable $1,000 per child tax credit. We go on to suggest a more comprehensive approach that integrates the individual income tax and Social Security tax systems into a single tax system with refundable $2,000 per working parent earned income tax credits and $1,000 or $2,000 per person refundable personal tax credits or universal grants. This paper considers how to make the federal tax system more work-friendly and less distortionary for low- and moderate-income workers.
Notes from this section of the paper
* This paper builds on Forman (2004).
1. Before tax credits are applied.
2. See, for example, Rosen (2002), pp. 375-84.
3. See, for example, Rosen (2002); Eissa (1996); Coe et al. (1998); Richards (1999); Triest (1996); CBO (1996); Danziger et al. (2002); and Shaviro (1997).
4. See, for example, Carasso, Rohaly, and Steuerle (2003, forthcoming); Sawicky, Cherry, and Denk (2002); Sawhill and Thomas (2001); and Ellwood and Liebman (2000).