Public policy towards low-income families with children has undergone a paradigm shift in the U.S. over the past two decades. Traditional welfare programs, such as Aid to Families with Dependent Children, relied on direct cash assistance to low-income families with children in an effort to increase income and reduce poverty among the disadvantaged. As epitomized by Hilary Hoynes in a 1997 piece, a major criticism of these public assistance programs was that they created significant work disincentives leading to long-term poverty, and also discouraged the formation of two-parent families.
Over time, therefore, we have seen a dismantling of these traditional programs and an increasing reliance on the tax system as a means of providing cash support for needy families. A series of tax acts starting with the 1986 Tax Reform Act have increased assistance to the working poor through programs such as the Earned Income Tax Credit (EITC), the Child and Dependent Care (CDC) credit, and the Making Work Pay credit among others.
In fact, currently, the tax code allows low-income individuals and families (at varying income levels) seven different tax credits (including the refundable and non-refundable portions of each credit). The credits are either tied to certain expenditures such as child care expenses and education expenses or are provided as incentives to low-income families who work. Each has varying income and other eligibility requirements, different schedules, different maximum credit values, and different phase-in and phase-out ranges, adding layers of complexity and high marginal tax rates even at the lower end of the income distribution.
The scale of these tax credit programs has increased exponentially in recent years. For instance, there has been a spike in the Child Tax Credit. The total credits have more than doubled between 2000 and 2006, rising from approximately $20 billion to about $47 billion. The refundable portion of the Child Tax Credit has grown by more than 1,000 percent, while the non-refundable portion has grown by about 60 percent.
The EITC has also grown by about 40 percent over this entire period, and the education credits by about 45 percent. On average, the total size of these government transfer programs has grown by about 70 percent, from $60 billion in 2000 to $103 billion in 2006.
How effective have these tax credit programs been in reducing poverty rates in the U.S.?
The Census Bureau reports several measures of poverty based on different measures of annual income. The official poverty rate is based on annual money income for the family (before taxes), while several alternative measures of poverty adjust money income for government taxes and transfers. The latter definition directly captures the effect of government programs on the poverty rate.
By both definitions, however, poverty levels have increased in recent times by nearly a percentage point. Between 2000 and 2006, the percent of families in poverty increased from 7.9 percent to 8.8 percent under Definition 18 of the alternative measure, and from 8.6 percent to 9.8 percent under the conventional measure. These numbers suggest that in each year, the net effect of government tax and transfer programs on poverty rates is beneficial. However, the trend over time is somewhat discouraging. Despite the tremendous growth in the size of the credits, there has been no significant dent in poverty rates using the alternative measure.
One possible factor is that these credits are underutilized. As noted by a 2005 summary of the literature by Dickert-Conlin, Fitzpatrick, and Hanson, studies suggest that participation in the EITC among those eligible to file is less than 100 percent. Matching CPS data to tax returns, the IRS estimates that approximately 64.2 percent of those eligible claimed the EITC in 1996. Using the self-reported data from the Survey of Income and Program Participation, the IRS also estimates an EITC filing rate of at least 73.5 percent. Other surveys show that knowledge of the EITC is negatively correlated with factors like low education, low income, and Hispanic ethnicity, suggesting that noncompliance is unintentional.
Literature on the CDC similarly finds that low-income filers do not use the credit, since it is not refundable. In a 1996 piece in the National Tax Journal, Altshuler and Schwartz use a ten-year panel of tax return data to show that taxpayers in the lowest decile and many in the second decile do not have a large enough tax liability to use the CDC. In their 1996 piece, Gentry and Hagy found that in 1989, less than three percent of families with dependents that made less than $10,000 claimed the credit, compared with 15.7 percent overall. Using 1998 tax return data, a 2003 article by Eiler and Hrung find that no taxpayers in the bottom two deciles benefit from the credit.
Again, the education credit is not heavily utilized by low-income taxpayers. Around half of all returns filing for an education credit have adjusted gross income above $40,000, and as noted by Dickert-Conlin, Fitzpatrick, and Hanson, more than half of the credit dollars accrue to this group. The GAO estimates that 40 percent of all college undergraduates received an education tax credit from 1999 to 2000, but for undergraduates with family income less than $20,000, that number dropped to four percent for dependents and nine percent for independent undergrads. According to Holitzell and Smith, of those that did not file for the credit, 59 percent believed they were ineligible, and 27 percent were unaware that the credit existed.
Hence, it is clear that each of these programs has a bewildering array of eligibility rules, with the result that some families that are entitled to these benefits do not file for them. Further, the current system complicates the marginal tax skyline, causing sudden spikes in tax rates as the credits phase out. Figure 1 shows the marginal tax schedule for all income earners starting from a low income of $0 to a high of $200,000 for the married couple and $100,000 for the single parent. The two charts look marginally different since the credits apply differently to married families and single filers.
Initially, as the credits, especially ones like the EITC, are phased in, the marginal tax rates are negative. The negative rates are a consequence of the fact that while taxpayers in this group have zero tax liability, they are still entitled to the refundable credits. In the phase-out region of the credits, the marginal tax rates start to rise, leading to the steep upward shift in the schedule. In fact, at about $22,000, the marginal tax rate jumps up to 21 percent from being negative at $21,000. Again, after the phase-out of the EITC, the marginal rate jumps down from about 40 percent to about 28 percent as taxpayers get onto the regular income tax schedule.
Apart from the complexity that the current system introduces to the tax code, another issue is that the credits are not properly targeted towards low-income people. Data from the IRS show that in 2006 (the latest year available), the government paid out about $25 billion to those with annual incomes above $50,000, and more than $4 billion to individuals in the $100,000-$200,000 income range. These numbers are likely to be even higher today. Further, these do not include the American Opportunity Tax Credit and the Make Work Pay Credit, which have been enacted in 2009. There are also other credits available to businesses, such as the Investment Tax Credit, that would raise the credit amounts at the upper end of the distribution.
In a recent Tax Notes article, co-authored with Lawrence Lindsey, we explore several alternatives that could tremendously simplify the current panoply of tax credits available to low-income people. Initially, we focused on credits that are provided as a lump sum to eligible individuals. Since under the current credits the maximum size of the credit is about $7,000-$7,500, we used similar amounts to see how many filers could benefit from such a system that does not rely on complex eligibility rules and does not have phase-in or phase-out ranges.
While the advantage of such a system is its simplicity, the biggest drawback is the existence of a cliff at the point where the individual receives a dollar of income that disqualifies him from the credit. With limited information, individuals facing these cliffs would face steep marginal tax hikes (in the thousands of percent) that would cause them to lose all of their benefits on the next dollar that they earn. Therefore, for practical reasons, such a policy is unlikely to be adopted.
A better alternative would be to use the current EITC design, but to fold all of the various credits into one simple credit. For these schemes, we phase-in at a high rate of, say, 60 percent, reach a maximum level of credit and then phase out at 22 percent. In other words, we return 60 cents on every additional dollar earned up until the maximum level of credit, and then the value of the credit reduces by 22 cents for every additional dollar earned. The 22 percent tax rate is chosen to match the phase-out of the EITC, the highest credit phase-out in the current tax code.
If we keep the rates the same for married and single filers, the cost is approximately $232 billion. If this policy is enacted and is further restricted only to families with children, the total cost of the policy would match the current cost of credits and would extend up until an income level of $100,000. We believe that this is the most practical policy and could ensure to some extent that the proposed new credit achieves the twin goals of distributional and revenue neutrality.
Our study reignites the debate on an issue that received substantial attention earlier in this decade with papers by Carasso, Rohaly, and Steuerle (2003); Sawhill and Thomas (2001); Sawicky, Cherry, and Denk (2002); Ellwood and Liebman (2000); and Steuerle (2000). Some of the most extensive and detailed work on this issue was conducted by the President’s Advisory Panel on Federal Tax Reform. While the specific proposals differ, the underlying objective of each of these was to consolidate and strengthen tax programs that benefit low-income people or, in the case of the Tax Panel, to substantially rationalize and simplify all aspects of the tax code.
As is clear from the data, the growth in tax credits has had little impact on poverty rates. The failure of the current credit system to better address poverty can be blamed both on targeting and a lack of awareness. A simplified credit system could solve both those problems.
Kevin Hassett is a senior fellow and director of economic policy studies at AEI. Aparna Mather is a research fellow at AEI.