Northwestern Journal of Law and Social Policy

I.    Introduction

¶ 1         This article examines participation in the Earned Income Tax Credit (EITC) and Temporary Assistance for Needy Families (TANF) programs among young adult (nineteen to twenty-five years of age), low-income families in 2004. Section II provides a brief overview of both programs and reviews related literature about program use. Section III poses the main study questions, describes the study methods, presents findings, and discusses the implications for social welfare policies. The study finds that program use varies by age, number of children in the household, ethnicity/race, work effort, marital status, prior program participation, and sex of participants. Those with two or more children, the least education, ethnic/racial minorities, women, and married persons are more likely than their respective counterparts to take advantage of either or both vis-a-vis neither program. Findings suggest that EITC and TANF are well-targeted, but underutilized programs.

II.    Program Overviews

¶ 2         The EITC is a major means-tested program¹ and the only one that is a refundable tax credit.² If the amount of the credit exceeds the taxpayer's federal income tax liability, the excess is payable to the taxpayer as a direct transfer payment.³ Enacted in 1975 during the Ford administration as part of Title I of the Tax Reduction Act⁴ to offset the burden of the Social Security tax on low-income working parents,⁵ the EITC, made permanent during the Carter administration,⁶ generally equals a specified percentage of wages up to a maximum dollar amount. The maximum amount applies over a certain range of income and diminishes to zero over a certain income range. The EITC thereby has three ranges: phase-in, maximum credit, and phase-out, each of which varies by the number of qualifying children in the family. In the phase-in range, the EITC acts as a wage subsidy—as the family earns more, the transfer increases. In the maximum credit range, the transfer remains constant regardless of earnings. In the phase-out range, the EITC acts like a negative income tax—as the family earns more, the transfer is reduced.⁷ While the phase-in range provides a work incentive, the maximum and phase-out ranges have work disincentives for some families.⁸

¶ 3         The income ranges and percentages have increased several times since 1975, expanding the credit, as have the number of participants. It should be noted that the congressional mindset favoring the EITC in the 1970s coalesced in the context of debates about unsuccessful welfare reform initiatives that had embodied pro-work, pro-growth, and low-cost policies.⁹ The EITC program matured as an anti-poverty strategy in 1986 when Congress raised the maximum benefit, which had fallen by thirty-five percent in real terms, to the 1975 level, increased the phase-out level to near the 1975 level, and most importantly guaranteed the future integrity of the EITC by indexing it for inflation.¹⁰ Nonetheless, a study of EITC participation and compliance concluded that households having a legal requirement to file income taxes were more likely to file for the EITC than those without such a filing status (such as families receiving public assistance), with EITC take-up rates somewhere in the range of thirty-one to thirty-nine percent in tax year 1988.¹¹ In 1991, adjustments were made for low-income workers with one or two or more children,¹² and in 1994, low-income workers with no children were eligible,¹³ increasing the take-up rate to an estimated fifty percent in 1999.¹⁴ Increases in the number of EITC claimants reflected these changes. In tax year 1975, the first year of the EITC, there were 6.2 million returns claiming the credit, staying constant until 1985, when 6.5 million claimed the credit, doubling to 12.5 million returns in 1990, and nearly tripling to 19.3 million in 1995.¹⁵ By 2005, the most recent year of available data at the time of this study, nearly 22.8 million returns were filed claiming the credit, nearly seventeen percent of all tax returns filed for that tax year.¹⁶

¶ 4         In 2008, the minimum income for the maximum credit was $5720 for individuals with no children, $8580 for those with one child, and $12,060 for those with two or more children; the phase-out income range began at $7160 for individuals with no children and $15,740 for those with children, and ended at $12,880 for individuals, $33,995 for those with one child, and $38,646 for those with two or more children.¹⁷ It should be noted that unlike public assistance programs in some states, single-parent and two-parent families with similar income levels receive the same EITC benefit, and two-parent families with similar income levels receive the same EITC benefit regardless of whether one or both parents work.¹⁸ The three benefit formulas (for no children, one child, and two or more children) have remained in place since 1994 and the subsidy rates have stabilized respectively at 7.65%, 34%, and 40% since 1997.¹⁹

¶ 5         Prior to 2002, low-income taxpayers with a "married filing separate" status were ineligible for the credit. This policy created a bonus for two very low-wage workers with children because their joint return entitled them to a higher credit than would have been the case had they filed separately, given the eligibility and phase-out levels of the credit at the time of this study.²⁰ However, a two-earner couple with children and $35,000 of combined income was ineligible for the EITC if married, but eligible for a sizable credit if they did not marry, live together, and raise a family.²¹ Although these two latter couples had similar income and family responsibilities, they were not treated the same under the tax code, violating the principle of horizontal equity.²² Many of the compliance problems that the Internal Revenue Service (IRS) faced, and that were addressed in mid-to-late 1990s legislation,²³ were a function of the same issue, namely the relative treatment of single and married taxpayers. Achieving marriage neutrality, progressiveness, and equal taxation of couples with the same income was (and remains) a longstanding tax problem.²⁴ Finally, it should also be kept in mind that twenty-three states and the District of Columbia have implemented their own EITC programs, adding between five and twenty-five percent more credit above what is reflected in the present assessment of the Federal initiative.²⁵

¶ 6         Enacted into law on August 22, 1996, the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) replaced the sixty-one year old program Aid to Families with Dependent Children (AFDC) with the block grant program Temporary Assistance for Needy Families (TANF), beginning July 1, 1997.²⁶ The AFDC program had survived two other reform efforts: the Nixon administration's Family Assistance Plan, which would have nationalized it and expanded cash benefits to low-income workers,²⁷ and the Family Support Act of 1988, which strengthened the link between welfare receipt and work while retaining the federal entitlement nature of the program.²⁸ PRWORA authorized states to formulate their own programs, but imposed work-trigger time limits, lifetime benefit limits, and minimum rates for work participation.²⁹ Enactment resulted in state-to-state variations in income thresholds for cash assistance eligibility³⁰ and the shortening of time limits below the federal five-year maximum for use of federal funds.³¹ PRWORA was silent concerning states' authority to decide whether to count EITC payments received by TANF recipients as income. The Taxpayer Relief Act of 1997, however, prohibited EITC payments to TANF recipients whose earnings were derived from workfare.³² The TANF program was renewed through September 30, 2010, as part of the Deficit Reduction Act of 2005.³³

¶ 7         The number of families TANF served declined from a high of 3.2 million in 1998, its first full year of operation, to 2.1 million in 2005.³⁴ Overwhelming majorities of TANF families were single-headed in 1998 and in 2005, about 99.9% in each year.³⁵ In 1998, there were 8.8 million recipients of TANF, 6.3 million (71.4%) of whom were children; in 2005, there were 5.1 million TANF recipients, 3.8 million (74.6%) of whom were children.³⁶ Take-up rates of the AFDC program hovered between seventy-eight to eighty-five percent throughout most of the 1980s and early 1990s, but began a precipitous decline after 1994, dropping to fifty-six percent of eligible families for TANF in 1998 and to forty-two percent in 2004.³⁷ As welfare participation decreased, as could be expected, work effort was also found to increase. One Michigan-based study of randomly selected single mothers obtained from the Women's Employment Survey showed an increase in the percent of the sample of those who worked twenty hours or more in during the week of being surveyed but received no welfare to 37.3% in 1998 from 19.9% in 1997.³⁸ The study also found declines in the percent of those who worked and received welfare to 24.2% in 1998 from 36.9% in 1997, and of those who did not work but received welfare to 23.8% in 1998 from 35.0% in 1997.³⁹

¶ 8         There was some evidence that participation in welfare programs increased the claim rate for EITC,⁴⁰ although the evidence from most studies suggested an inverse correlation between the two programs—that is, as participation rates in TANF declined, those in EITC increased. In one study, counties with a hundred percent enrollment rate in their respective California's Welfare-to-Work programs vis-a-vis counties where nobody enrolled in such programs were found to have increased the EITC claim rate by 2.3 percentage points.⁴¹ However, in one of the first studies to provide direct estimates of the EITC on welfare use, Jeffrey Grogger reported that the EITC contributed to decreased use of welfare among female-headed households between 1978 and 1999.⁴² This was especially the case for families with the oldest among the youngest children: "[A] $1000 increase in the maximum credit would reduce welfare use by an average of 1.5 percentage points among families whose youngest child was three and by 3.1 percentage points among families whose youngest child was ten."⁴³ A study that relied on female TANF recipients in Wisconsin in 1997 provided indirect support for Grogger, reporting that TANF participation rates in 1999 and 2000 were twenty-nine percent, whereas EITC participation rates were seventy-five percent.⁴⁴

III.    Study

¶ 9         The present study relied on panel data to examine the relationship between EITC and TANF participation among self-reported EITC-eligible low-income families. It addressed the following questions:

(1)    What proportion of EITC-eligible low-income families participated in neither EITC nor TANF?

(2)    What proportion of EITC-eligible low-income families participated only in EITC?

(3)    What proportion of EITC-eligible low-income families participated in both EITC and TANF?

(4)    What proportion of EITC-eligible low-income families participated only in TANF?

(5)    How did program participation among EITC-eligible low-income families vary by sociodemographic characteristics such as education, ethnicity/race, marital status, and sex?

Answers to these questions were meant to identify gaps in program use such that outreach efforts might be better targeted to increase take-up rates.

¶ 10         Data came from Round 9 of the National Longitudinal Survey of Youth 1997 (NLSY97), the newest survey in the NLS program and the most recent round of data available at the time of this study.⁴⁵ NLSY97 was designed to be representative of people living in the United States in 1997 who were born between 1980 and 1984.⁴⁶ Many of the oldest youth (age sixteen as of December 31, 1996) were still in school at the time of the first survey and the youngest respondents (age twelve) had not yet entered the labor market. The NLSY97 cohort included 8984 individuals. Responsibility for administration of the NLS resides with the U.S. Department of Labor, Bureau of Labor Statistics, which contracts with both the Center for Human Resource Research at The Ohio State University and the National Opinion Research Center (NORC) at the University of Chicago to manage the NLS program, share in the design of the survey instruments, disseminate the data, and interview respondents.⁴⁷ The Round 9 sample included 7338 individuals, a retention rate of 81.7% in 2005.⁴⁸ Differences between the 1997 Round 1 and 2005 Round 9 respondents were found by education and ethnicity/race in 1997: the Round 9 sample had completed fewer years of schooling at the time of the first interview in 1997 than had the Round 1 sample (7.65 versus 7.96, p < .001) and white respondents had the lowest retention rate, 80.0%, compared to 84.8% for black and 82.0% for Hispanic respondents (p < .001). Though statistically significant, these differences were deemed slight, but were nonetheless kept in mind for analysis and discussion purposes. No difference was found on families' 1997 income-to-poverty ratios between Round 9 and Round 1 respondents (2.92 versus 2.81). The study sample comprised 1098 individuals nineteen to twenty-five years of age whose self-reported family income in survey year 2005 for calendar year 2004 was at or below twice the federal poverty thresholds for that calendar year and for whom information on all other study measures were available. Respondents who self-reported that they were not eligible for the EITC were excluded from the study sample.

¶ 11         Program participation status (PPS) in 2004 was the focal variable of the study. Respondents were classified in the following mutually exclusive categories: EITC participants, TANF participants, EITC and TANF participants, or neither EITC nor TANF participants. EITC participation was derived from a survey item asking respondents whether they (or their spouses or partners) claimed or planned to claim an Earned Income Tax Credit on their 2004 federal income tax return.⁴⁹ Those who responded that "yes," they (or their spouses or partners) did claim or planned to claim, were classified as EITC participants for purposes of this study. Those who responded "no," because they were either unaware of the EITC, or for some other reason, were deemed not EITC participants and classified accordingly. TANF participation was derived from a series of looped items asking respondents whether they, their spouses or partners, or their children were receiving payments for low-income families and if so, the month and year of payment receipt.⁵⁰ Respondents who reported TANF payments for any month in 2004 were classified as TANF recipients.⁵¹ EITC and TANF program participation measures for 2003 were also created from similarly worded items on prior surveys.⁵² EITC participants in 2003 were coded as 1, nonparticipants as 2; TANF recipients in 2003 were coded as 1, nonparticipants as 2.

¶ 12         Staff at the Center for Human Resource Research (CHRR) at The Ohio State University created several study measures. Socioeconomic status (SES) was obtained from the income-to-poverty ratio (IPR), which was determined by dividing household income by the household poverty threshold level, accounting for household size and source of income of each respondent.⁵³ An IPR less than or equal to one signified that respondents lived in households whose income was at or below the poverty threshold. These respondents were classified as poor. As previously noted, respondents reporting an IPR more than twice the federal poverty thresholds were excluded from the study. Hence, SES comprised two groups, the poor (IPR ≤ 1) coded as 1 and the near-poor (1 > IPR ≤ 2) coded as 0. Hours worked was created for each calendar year. Respondents not working in a given year were given a default value of zero (0) hours worked.⁵⁴ Respondents who worked less than 1000 hours in 2004 were coded as 1, while those who reported working more than 1000 hours were coded as 0. CHRR staff also created the measure number of children under eighteen years of age in the household.⁵⁵ Respondents living in households with more than one child under the age of eighteen were coded as 1, others as 0.

¶ 13         Sociodemographic measures included age, education, ethnicity/race, marital status, and sex. Age was reported at the time of interview. Respondents twenty-one years of age or younger in 2004 were coded as 1, others as 0. Education was measured as highest grade completed in 2004: Respondents completing fewer than twelve years of education were classified as less than high school; those completing twelve years of schooling were classified as high school graduates; those with more than twelve years of completed schooling were classified as beyond high school. Ethnicity/race comprised the mutually exclusive categories of Hispanic, black (non-Hispanic), and white (non-Hispanic). Marital status was coded such that 1 = married, 0 = unmarried. Sex was coded such that 1 = female, 0 = male.

¶ 14         Multinomial logistic regression was used to examine the capacity of study measures to predict PPS. The reference category was neither EITC nor TANF participants. All analyses were done using SPSS 15.0.

¶ 15         As Table 1 shows, more than half of the sample (51.0% weighted) participated in neither EITC nor TANF. Using 55% as the lower-limit cutoff, least likely program participants included younger young adults (56.4%), young adults with one or no children (57.0%), the most educated (57.8%), unmarried persons (55.8%), non-EITC participants in 2003 (64.4%), those with family incomes below poverty (57.3%), and males (55.2%). Using 45% as the upper-limit cutoff, most likely program participants were TANF participants in 2003 (9.8%), married persons (22.5%), EITC participants in 2003 (27.0%), those with two or more children (28.1%), those with less than a high school education (35.8%), and Hispanics (39.3%).

¶ 16         Slightly fewer than half (46.8% weighted) of the young adults participated only in EITC, 2.2% participated in both EITC and TANF, and 1% only in TANF. Older young adults were more likely to participate in these programs than not (54.8% versus 43.6%), with 50% of those 22 years of age or older participating in EITC only versus 42.5% of those 21 years of age or younger. Nearly two-thirds (63.4%) of young adults with two or more children participated in EITC, while more than two-fifths (41.5%) of those with one or no children participated in EITC only. The least educated young adults were more likely than the most educated to participate in EITC only (58.0% versus 40.6%), TANF only (2.9% versus 0.5%), and both EITC and TANF (3.3% versus 1.1%). White young adults were the most likely to participate in neither program (51.4%) versus black (42.1%) and Hispanic (39.3%) young adults. Black young adults were more likely to participate in TANF only (2.4%) than were Hispanic (1.3%) or white (0.6%) young adults; Hispanic and black young adults, however, were as likely as each other to participate in EITC only (54.5% and 53.5%, respectively) and were more likely to do so than white (43.5%) young adults.

¶ 17         As could be expected, young adults who worked more than 1000 hours in 2004 were more likely than those who worked fewer hours to participate in EITC only (49.6% versus 42.1%) and less likely to participate in TANF only (0.6% versus 1.7%), but they were also less likely to participate in neither program (47.5% versus 54.2%). Nearly three-fourths (73.0%) of married young adults participated in EITC only versus only 41.2% of unmarried young adults, while unmarried young adults were twice as likely than married young adults to participate in TANF only (1.1% versus 0.5%). Young adults who had participated in EITC in 2003 were more likely to participate in 2004 (68.9%) than were those who had not participated in EITC in 2003 (32.9%), but they were almost as likely to participate in TANF only (1.1% versus 0.9%). Young adults who had participated in TANF in 2003 were more likely to participate in 2004 (16.8%) than those who had not participated in TANF in 2003 (0.4%), and they were less likely to participate in EITC only (30.4% versus 47.4%).

¶ 18         As also could be expected, near-poor young adults were more likely to participate in EITC only than were poor young adults (52.8% versus 39.0%). More than three-fifths (63.1% not shown in table) of near poor young adults worked more than 1000 hours per week versus 45.0% of poor young adults. Near poor young adults were also more likely than poor young adults to participate in TANF only (1.4% versus 0.7%) and almost as likely to participate in both EITC and TANF (2.1% versus 2.3%). Women had slightly greater participation percentages than men for EITC only (49.9% versus 42.7%), TANF only (1.3% versus 0.7%), and both EITC and TANF (2.8% versus 1.4%).

Note: The referent category for the multinomial model is "Neither EITC nor TANF"; for Education, "Beyond High School"; for Ethnicity/race, "White." ***p < .001, **p < .01, *p < .05

¶ 19         As Table 2 shows, Hispanics and blacks were 1.5 (p < .05) and 1.6 (p < .01) times respectively more likely than whites to participate in EITC only in 2004 than in neither EITC nor TANF. Those who worked more than 1000 hours during 2004 were also 1.6 (p < .01) times more likely than those working less than that to participate in EITC only than in neither EITC nor TANF. Unmarried persons were 3.4 (p < .001) times (1/.296) less likely to participate in EITC only. Those who had participated in the EITC program in 2003 were 3.9 (p < .001) times more likely than those who had not to participate in EITC only in 2004. Women were 1.4 times more likely than men to participate in EITC only in 2004 than in neither EITC nor TANF.

¶ 20         As Table 2 also shows, the only statistically significant predictor of TANF only participation in 2004 was participation in TANF in 2003. Prior TANF participants were more than 200 (p < .001) times more likely than non-participants of TANF in 2003 to participate in TANF only in 2004 than in neither EITC nor TANF.

¶ 21         Finally as Table 2 reflects, those twenty-one years of age or younger were 10.6 (p < .001) times more likely than older persons to participate in both EITC and TANF than in neither program. Those with two or more children in the household were 3.8 (p < .05) times less likely to participate in both EITC and TANF. Prior TANF participants were more than 378 (p < .001) times more likely than non-participants of TANF in 2003 to participate in both EITC and TANF than in neither program. Women were 4.5 (p < .05) times more likely than men to participate in both EITC and TANF.

¶ 22         Findings of the study suggest that EITC and TANF are well-targeted, but perhaps that they are also underutilized as a whole. Program use (EITC only, TANF only, or both EITC and TANF) varies by age, number of children in the household, ethnicity/race, work effort, marital status, prior program participation, and sex of participants. Those with two or more children, the least education, ethnic/racial minorities, and women are more likely than their respective higher earning counterparts to take advantage of either or both programs.⁵⁶ It is in this sense that EITC and TANF are considered well-targeted.⁵⁷ The low participation rates of TANF—less than five percent even among those whose family income fell below official federal poverty thresholds—might be understandable given that reduced welfare dependency was a goal of the 1996 legislation. It would not be unreasonable to speculate that the imposed maximum five-year lifetime limit of federal cash assistance for program participants has a discouraging effect on low-income young adults. More research needs to be done to discern the extent to which time limits function to discourage TANF participation. How do the take-up rates for TANF compare to those of the Aid to Families with Dependent Children program? To the extent TANF take-up rates are found to be less than those of AFDC, what factors, other than perhaps the effects of sample selection, as seemed to be the case in this study, might account for these rates? Are low-income individuals hedging their bets by delaying use at relatively young ages so that they remain TANF eligible should equal or more dire need arise in the future? To what extent do state application procedures and/or compliance strictures discourage TANF use? Given the five-year federal time limit and even more restrictive time limits set by many states,⁵⁸ perhaps the more perplexing question is whether advocates for low-income persons and their families should seek to increase the percentage of TANF participation among low-income young adults.

¶ 23         The relatively low participation rate for EITC found in this study is consistent with previously mentioned studies showing take-up rates in the vicinity of fifty percent,⁵⁹ although it is higher than the IRS estimates of twenty to twenty-five percent for 2006 reported by Treasury Secretary Henry Paulson on January 31, 2008.⁶⁰ The IRS seems to be of two minds in regard to the EITC take-up rate. On the one hand, its five-point initiative to improve the service, fairness, and compliance in the administration of the EITC focuses more on quality control to minimize abuse, especially in regard to qualified children residing in the household.⁶¹ On the other hand, the IRS launched campaigns to help low-income taxpayers to take advantage of the EITC.⁶²

¶ 24         Several states have their own EITC outreach initiatives and related websites, such as Michigan,⁶³ North Carolina,⁶⁴ West Virginia,⁶⁵ as well as cities such as Indianapolis,⁶⁶ Nashville,⁶⁷ Manchester, CT,⁶⁸ and New York.⁶⁹ Corporate America has also responded with websites dedicated to promoting use of the EITC,⁷⁰ as has the non-profit sector.⁷¹ The IRS website proclaims that it is easier than ever to find out if one qualifies for the EITC, and it links to related information in English and Spanish.⁷² There is even a national EITC outreach initiative promoted by the Center on Budget and Policy Priorities.⁷³ This initiative lists a number of national web-based resources,⁷⁴ including the National Tax Coalition⁷⁵ and the Hatcher Group State EITC Online Resource Center.⁷⁶

¶ 25         Findings of this study suggest that low-income groups in general can benefit from greater outreach. Paradoxically, given the overall well-targeted nature of the program, findings also suggest that "less needy" groups such as those who go beyond high school, non-Hispanic whites, and males could also benefit from greater or better targeted outreach efforts, especially in regard to EITC. Findings also suggest, however, that the "most needy" persons, that is those whose family incomes fall below poverty thresholds, are the least likely to use these programs and would benefit from even greater outreach efforts. Given that poor families are less able to afford computers in their homes and hence have less access to the internet to take advantage of web-related EITC outreach sites, social service agencies and legal aid offices would do well if they were to have financial counselors on their staff who could take extra steps when necessary to ensure that those low-income working families below the poverty thresholds avail themselves of the EITC. They might, for example, make dedicated computer time available to such clients between January and April for the specific purpose of assessing EITC eligibility and assisting with tax preparation when warranted. They might also establish formal relationships with local tax preparation agencies and refer potentially EITC eligible clients accordingly or with local libraries that provide technology access for individuals in poverty.⁷⁷

ENDNOTES