Publications

In this article, we look at how benefits cliffs differentially affect minority populations. We find that:

• Safety net programs provide critical financial support to low-income families, but the way these programs are structured may pose a barrier to workers who want to advance in their careers and disproportionally burden Black and Hispanic families.
• Even if workers advance in their careers in spite of the existing safety net program structure, they may not be financially better off than before career advancement, which raises concerns that workforce development programs—despite their best intentions—are not increasing a worker's standard of living but only moving the worker off public benefits.
• The implications of safety net program design vary across the general population and across race and ethnicity, largely because of historic and existing disparities in income, which may affect the ability of workforce development approaches to reduce racial income and wealth disparities.

Read on to learn more about this issue and possible policy and practice interventions to reach desired outcomes.

Social safety net programs and tax rates
Institutions across the United States are examining policies and practices to address these long-standing racial inequities. The goal is a more inclusive society and a more inclusive economy. Improving access to high-quality job training and career advancement opportunities is often discussed as a promising strategy to help reduce racial inequalities in income and wealth. For these workforce strategies to succeed, career advancement with accompanying income gains—through the acquisition of new skills or the transfer of existing skills to a high-paying occupation—have to improve the worker's standard of living. In some cases, however, income gains may not result in an increased standard of living.

Social safety net programs—for example, food, childcare, and housing assistance—provide critical financial support to low-income families and are necessary to enable individuals to meet their basic needs while they are pursuing training and education to advance in their careers. Therefore, we have been studying these programs to understand how they, in tandem with workforce development efforts, improve a worker's standard of living, facilitate economic self-sufficiency, and address racial wealth and income inequalities.

Most of the major federal safety programs such as food assistance, childcare assistance, and housing assistance are means-tested (that is, only individuals with income below certain thresholds are qualified). As low-income workers advance in their careers and earn more money, the amount of support from safety net programs decreases. This loss of means-tested safety net support is an effective marginal tax rate (EMTR) on income gains. High effective marginal tax rates mean that some workers have a financial disincentive to invest in their own human capital and advance from lower-wage work to jobs that lead to economic self-sufficiency.²

An extensive body of research examines EMTRs for different family compositions, but surprisingly little work examines how EMTRs vary across racial/ethnic demographic groups.³ To the extent that EMTRs create disincentives for career advancement or lead to smaller gains in a worker's standard of living, racial discrepancies in EMTRs may exacerbate the wage and wealth disparities that exist as an enduring result of a history of discriminatory policies.⁴

EMTRs may be higher for racial and ethnic minorities than whites for several related reasons. First, the distribution of income and wealth can vary by race, and the safety net programs that lead to higher EMTRs are targeted to workers with lower incomes. For instance, within the general population, EMTRs are highest for the bottom half of the wealth distribution. Since non-Hispanic Black and Hispanic families have, on average, lower household wealth and income than Asians and whites, they are overrepresented in these higher EMTR brackets. These income differences alone would not lead to differences in EMTRs, but income differences combined with the structure of the safety net program can lead to these EMTR differences. Racial and ethnic minorities are more likely than whites to participate in safety net programs, with participation defined as the share of the group accessing a safety net program. A 2019 Urban Institute report analyzed 2012–14 data, the most current years available at the time of its publication. The report found that 36 percent of non-Hispanic Blacks and 28 percent of Hispanics, on average, participated monthly in at least one major safety net program, compared to 13 percent of non-Hispanics whites, even though whites make up the largest share of total participants.⁵

Second, family composition can vary by race. Means-tested programs offer greater financial support to families with children. Though there is regional variation, nationally, Black and Hispanic families have, on average, more children under the age of 18 than Asian and white families.⁶

Third, the geographic distribution of the population across the United States can vary by race. Benefits rules and availability⁷ differ widely across states, which may affect the racial and ethnic differences in EMTRs to the extent the group population distribution correlates with the public benefits system structure across states.

Fourth, institutional and behavioral factors that affect take-up rates vary across groups. It is possible for the share of a group eligible for a program to be equal across groups, while the percentage of those eligible who take up the program varies by group. For example, awareness of the program, the ease of applying for the program, the stigma associated with certain programs, and the capacity to comply with rules may vary across groups, leading to differential take-up rates.

Our analysis examines a subset of the first two reasons—income differences and family composition. We will address the other listed reasons in future work. Thus, the scope of this article is limited to examining how safety net program structure, combined with incomes and family types that vary by racial and ethnic groups, can create racial disparities in EMTRs. In our analysis, we find that the first reason—specifically, racial and ethnic differences in income—explain most of the variation in EMTRs.

Our analysis builds on increasing attention to how the U.S. tax system disproportionately aids whites compared to people of color. For example, a recent project by the Tax Policy Center and the Urban Institute uses a hypothetical 1040 form to illustrate the numerous ways in which the tax code may negatively affect racial minorities. One example is preferential tax treatment for capital gains, which benefit white families because they are more likely than Black and Hispanic families to hold financial assets taxable as capital gains. This informative analysis makes an important point: overall, the U.S. tax system is progressive, which could help reduce racial disparities in income and wealth, but some policies—like the preferential treatment for capital gains—may exacerbate inequalities. We argue that while means-tested programs redistribute resources to families in need and provide an important social safety net, they also create regressive marginal tax rates. Racial disparities in EMTRs raises further concerns about how current tax and transfer programs exacerbate racial disparities in income and wealth.

Methodology used
For each family in the 2018 American Community Survey (ACS), we calculate EMTRs based on family income, family composition (the number of adults and the number of kids of different ages), and state of residency.⁸ In our calculations of EMTRs, we include all major federal and state taxes and transfer programs. A detailed list of programs that are included in our methodology is in an appendix of this discussion paper. To estimate EMTRs, we increase the income of each family in the ACS by 10 percent. We then compute how the sum of the family's income and benefits change relative to this 10 percent increase in income. One minus this amount is the EMTR.⁹

We do not observe actual benefits program participation in the data so we cannot estimate EMTRs based on families' actual program participation. Instead, we calculate EMTRs based on two assumptions of benefit program participation.

First, we report EMTRs assuming that a family receives the following federal government benefits for which they are eligible: the Supplemental Nutrition Assistance Program (SNAP), the Earned Income Tax Credit (EITC), the Child Tax Credit (CTC), Medicaid/Children Health Insurance Program (CHIP), Affordable Care Act [ACA] subsidies, housing subsidies, and childcare subsidies (CCDF). This assumption represents an upper bound on EMTRs for families, in which families are able to access all these major public benefit programs for financial support on necessities such as childcare, health care, housing, and food.¹⁰

Not all families who are eligible for these benefits receive them. For example, Nina Chien of the U.S. Department of Health and Human Services (HHS) estimates that in 2016 only 15 percent of children eligible under federal rules received CCDF subsidies. There are several reasons why some eligible families do not receive benefits. Demand for programs such as CCDF and housing subsidies often exceeds supply, resulting in lengthy wait lists. In addition, eligible families may not participate due to lack of awareness or confusion over the application and compliance procedures, as discussed in the literature on EITC participation. For this reason, we report EMTRs under the alternative assumption that the family receives the same set of federal government benefit programs except for housing and childcare assistance. Chien and Suzanne Macartney of HHS report that the most common benefit combination for households with children below 200 percent of poverty is SNAP, the EITC, the CTC, and Medicaid/CHIP. Thus, our assumptions for the "partial program participation" shown in this analysis approximate Chien and Macartney's modal low-income family on public benefits, except we also include Temporary Assistance for Needy Families.

In the appendix, we explore to what extent differences in income by race and ethnicity explain differences in EMTRs by race and ethnicity. To do this, we calculate EMTRs by race and ethnicity conditional on the families' position in the overall population income distribution. In other words, given similar incomes, to what extent do EMTRs vary by race and ethnicity?

Our findings
Table 1 shows results based on the participation in all benefit programs. For each race/ethnicity group, the table shows the distribution of EMTRs, from the 10th percentile of EMTR to the 90th percentile. Black and Hispanic families face higher median EMTRs than Asian and white families. Hispanic families have the highest median EMTR at 73 percent on the hypothetical 10 percent income increase. The 90th percentile of EMTRs is high for all race/ethnicity groups. White families have the highest 90th percentile EMTR: 10 percent of white families have EMTRs higher than 133 percent. Black families (128 percent) and Hispanic families (127 percent) have similarly high 90th percentile EMTRs, while Asian families (117 percent) have the lowest 90th percentile EMTR. Families with these EMTRs above 100 percent face so-called benefits cliffs: due to the loss of public benefits, the families are financially worse off after the earnings increase.

Appendix table 1 shows median EMTRs by race and ethnicity conditional on income. Within income quintiles, the differences in EMTRs by race and ethnicity are considerably smaller. This result suggests that racial and ethnic differences in income explain most of the variation in EMTRs.

Note: Sample includes families where the householder is prime working age (25–54) and income is greater than $10,000. White, Black, and Asian exclude those of Hispanic ethnicity.
Source: authors' calculations of American Community Survey data

In table 2, we show the distributions of EMTRs for the modal combination of public benefits receipt (TANF, SNAP, the EITC, the CTC, and Medicaid/CHIP only). EMTRs are lower than in table 1, which we expect since the analysis excludes valuable childcare and housing subsidies. We again find differences across race/ethnicity, but these differences are smaller than in table 1. Median EMTRs are highest for Hispanic families (48 percent) and Black families (47 percent) and lowest for white (44 percent) and Asian (41 percent) families. EMTRs at the 90th percentile are highest for Hispanic (73 percent) and Black (70 percent) families and lowest for white (66 percent) and Asian (61 percent) families. Even in the modal benefits case, Black and Hispanic families face higher median EMTRs and higher EMTRs at the upper end of the distribution, which disproportionally places a financial barrier to career advancement on Black and Hispanic families.

Appendix table 2 shows median EMTRs by race and ethnicity conditional on income for the partial benefits case. Similar to the full benefits case, we find that, within income quintiles, the differences in EMTRs by race and ethnicity are considerably smaller. Again, the result suggests that racial and ethnic differences in income explain most of the variation in EMTRs.

Note: Sample includes families where the householder is prime working age (25–54) and income is greater than $10,000. White, Black, and Asian exclude those of Hispanic ethnicity.
Source: authors' calculations of American Community Survey data

These national estimates show considerable variation in EMTRs across race and ethnicity. As part of this research project, we are examining how median EMTRs vary across the U.S. states and the District of Columbia. Several factors would explain cross-state variation, including the size of the racial income differential, differences in the number of children across race/ethnicity within state, and state laws. We will explore these differences in forthcoming work.

Recommendations for policy and practice
We conclude with these recommendations.

• To mitigate the high effective marginal tax rates, workforce development programs could require strategies that financially support workers as they enter training, complete training, and start work after training when they are likely to face benefits cliffs. For example, the Families Ascent to Economic Security (FATES) program, described in this paper, integrates a graduated phaseout of the subsidized childcare benefit into a career pathways workforce development program.
• Policy proposals include a set of strategies to smooth benefits cliffs (such as phaseout of programs or changing asset limits), case management that is aware of potential benefits losses and is able to coordinate access to financial resources in the community, career counseling that is aware of benefits cliffs and coaches workers on career paths to self-sufficiency, and work with employers to identify changes within their human resource departments that may mitigate benefits cliffs for lower-wage employees.
• Policymakers and program administrators could consider the disparate impact of high EMTRs on people of color when designing workforce development programs or targeted strategies to mitigate benefits cliffs. Exploring approaches such as targeted universalism could provide a road map to solutions.

By Elias Ilin, Atlanta Fed research associate and a PhD candidate at Boston University, Alexander Ruder, principal CED adviser, and Ellyn Terry, Atlanta Fed economic policy analyst specialist and a PhD candidate at the University of Washington

Note: Sample includes families where the householder is prime working age (25–54) and income is greater than $10,000. White, Black, and Asian exclude those of Hispanic ethnicity.
Source: authors' calculations of American Community Survey data

Note: Sample includes families where the householder is prime working age (25–54) and income is greater than $10,000. White, Black, and Asian exclude those of Hispanic ethnicity.
Source: authors' calculations of American Community Survey data

_______________________________________

¹ For example, see Bertrand and Mullainathan (2004).

² We define economic self-sufficiency as the amount of income needed for a worker to pay basic expenses as a minimally adequate level. See selfsufficiencystandard.org for more information on this concept.

³ For a review of this literature, see Altig et al. 2020.

⁴ See Hardy, Logan, and Parman (2018) for a discussion of individualist versus structural theories of poverty in the context of racial inequality, with structural theories emphasizing the long history of discrimination in labor markets, schools, housing, and neighborhoods.

⁵ EMTRs for a given racial and ethnic group will be affected by the share of that racial and ethnic group that participates in safety net programs, not the share of total participations in the population. For additional analysis of safety net participation by race and ethnicity, see here.

⁶ U.S. Census Bureau Current Population Survey, 2019 Annual Social and Economic Supplement, Table AVG3 available here. The average number of people under 18 in families is 1.96 for White alone, 2.12 for Black alone, 1.63 for Asian alone, and 2.14 for Hispanic.

⁷ For example, the length of waitlists for housing and childcare subsidies varies across states, metropolitan areas, and even counties.

⁸ We use data from the 2018 American Community Survey (ACS), one-year Public Use Microdata Sample (PUMS). The 2018 ACS is a 1-in-100 national weighted random sample (2,143 thousand housing units) of the U.S. population that provides detailed individual and household level social, economic, housing, and demographic data.

⁹ We additionally take into account any differences in health care premiums that results from changing health insurance due to changes in program participation (Medicaid has a $0 premium in most states, while the premium on Affordable Care Act [ACA] copay is roughly 30 percent of adjusted income).

¹⁰ A direct way structural racism in the legal system contributes to safety net access is limitations on safety net participation for felons. Compared to whites, Blacks are five times more likely to be incarcerated and Hispanics are twice as likely. See https://www.prisonpolicy.org/reports/rates.html.