Health Effects of the 2021 Earned Income Tax Credit Expansion on Young Adults Without Children

Introduction

One in five young adults ages 18–24 years experience poverty.¹ This percentage is higher than most other age groups and has been rising, while poverty rates for other age groups have declined.² Young adults are particularly vulnerable to poverty because of their lower earnings and reduced labor market opportunities compared to older adults who are considered more established in the workforce.^2–4 In addition, many US social safety net programs that lift other age groups out of poverty have historically excluded young adults at this key juncture of the life course. For example, young adults (<25 years old) without children are not eligible for the earned income tax credit (EITC). The EITC is the largest US poverty alleviation tax credit program for low- and moderate-income workers, providing $60 billion in tax refunds each year to about 25 million working families, disbursed as a lump-sum tax refund in the spring.⁵ Young adults without children are also ineligible for the Child Tax Credit (CTC), which by definition is only for adults with dependent children. Furthermore, although they are not categorically ineligible for the Supplemental Nutrition Assistance Program (SNAP), they receive smaller SNAP benefit amounts because of household size and strict work requirement eligibility.^6–8 Additionally, young adults without children are also less likely to receive benefits from the Temporary Assistance for Needy Families (TANF) program, constituting only 4% of TANF beneficiaries.⁹

The COVID-19 pandemic worsened labor market outcomes for young adults, with the unemployment rate for 16- to 24-year-olds reaching 24.2% at the beginning of the pandemic (in the second quarter of 2020), a jump of 15.9 percentage points from before the pandemic.¹⁰ Mental health has also worsened in recent years for young adults, partially fueled by loss of employment and income and because of pandemic-era stay-at-home orders and associated social isolation and loneliness.^11–16 Furthermore, widening income disparities in recent years have likely exacerbated existing health disparities, as low-income individuals have reduced access to nutritious food and opportunities for exercise, lack of access to healthcare, and resulting shorter life expectancy.^17–19

In 2021, Congress passed the American Rescue Plan Act, which extended eligibility for the EITC to young adults aged 19–25 years old without children (known as “childless adults”), who had been previously ineligible. The legislation also increased the credit amount for childless adults more generally from $500 to $1,500, and raised the income eligibility threshold from $16,000 to $21,000.²⁰ The expansion benefited more than 17 million childless adults, with about 11 million receiving the EITC for the first time in 2021.²¹ However, the expansion expired after one year. The EITC program has been one of the most studied US anti-poverty programs, implemented in the 1970s to encourage work while reducing poverty.²² As of 2023, 28 states also have their own state EITC programs.^5,23

In rigorous quasi-experimental and experimental studies on families with children, the EITC has generally been associated with positive health impacts, since it addresses poverty as a root cause of health disparities. Studies have found reduced psychological distress and depression and increased self-reported happiness.^24,25 In households with children, the EITC improves perinatal health, child health, and food security.^26–29 Despite concerns that EITC income benefits may increase the use of unhealthy substances, there have been no impacts on drug or alcohol mortality.³⁰ Importantly, studies evaluating the impacts of the EITC have focused primarily on families with children, since these were historically the primary beneficiaries of the policy. A handful of studies on the Paycheck Plus Program, which was an RCT of an expansion of the EITC to childless adults in New York and Atlanta, had mixed results, with physical and mental health benefits in New York but null results in Atlanta.^31,32 However, these studies were conducted before the COVID-19 pandemic (2017–2019), and evaluated a different expansion to childless adults. Moreover, the RCT paycheck plus study was limited to only a few locations (e.g., New York and Atlanta). Only two studies to our knowledge have evaluated the impact of the 2021 temporary expansion of the childless EITC. One study found it reduced material hardship and food insufficiency, while the other found improvement in mental health among young adult renters.^33,34

We aim to fill this gap in the literature, providing some of the first evidence on the health impacts of the 2021 temporary EITC expansion for childless young adults. We use a large national data set and a rigorous quasi-experimental design to test the hypothesis that the temporary income boost improved physical and mental health and health behaviors among newly eligible recipients. This research has important policy implications as state governments in particular continue to consider various expansions of the EITC.

Methods

Data and Sample

Data were drawn from the 2021–2023 waves of the Behavioral Risk Factor Surveillance System (BRFSS), a national serial cross-sectional survey of adult health and health-related behaviors conducted by all 50 states and US territories.³⁵ We limited the sample to young adults without children age 18–34 years with incomes below $50,000. We excluded respondents from US territories, as they are ineligible for the EITC. Those aged 18–24 represented the “treated” group who were previously ineligible for EITC benefits and became newly eligible during the 2021 tax year, while those aged 25–34 represented the “control” group of similar individuals who had already been eligible in previous years. Due to BRFSS age categories, our treatment group includes 18-year-olds which is slightly different from the EITC policy eligibility criteria of 19–24 years. We excluded those with children, since they were already previously eligible, and because they would have been separately impacted by other key policy changes during this time like the 2021 CTC expansion. We excluded those with higher incomes as they are likely to differ in important ways from low-income individuals. We excluded those interviewed in the months of July-December as they would have unlikely received the EITC refunds during those months. Finally, we excluded those who were unemployed because the EITC requires individuals to have earned income greater than $0, resulting in a sample size of 11,137. Supplemental Figure 1 provides the sample selection flowchart including the sample size of the study outcomes.

Outcomes

Outcomes included measures of mental health, physical health, and health behaviors. Mental health was captured by the number of poor mental health days in the past 30 days. We also calculated a binary mental health measure for whether individuals had fourteen or more bad mental health days in the past 30 days, a standard dichotomization of this outcome referred to as frequent mental distress.^36,37 Physical health was captured by the number of poor physical health days in the past 30 days. Binge drinking was captured with a binary variable for consumption of four or more drinks (for women) or five or more drinks (for men) on one or more occasions within the past 30 days, the standard definition.^38,39

Exposure

The EITC expansion was implemented in tax year 2021, which means individuals received the tax credits in the spring of 2022 when they filed taxes. Thus, we considered young adults interviewed in 2022 to be exposed to the policy. We restricted our analysis to February through June of each year, as recipients typically receive their EITC benefits during this time; this analysis therefore captures short-term impacts of the expansion.^40,41 The expanded EITC was not renewed, and it thus only affected individuals in tax year 2021 (survey year 2022). Those interviewed in 2021 and 2023 (representing tax years 2020 and 2022, respectively) were therefore considered unexposed.

Covariates

Covariates included variables that might potentially confound the relationship between the policy and the outcomes: race/ethnicity, education, sex, marital status, income, residence in a metropolitan versus non-metropolitan area, and health insurance coverage. We controlled for state fixed effects to account for state-level differences in the outcomes of interest. Race/ethnicity was self-reported, and we categorized it as non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic other. The latter group is heterogeneous, but small cell sizes precluded more granular analyses.

Difference-in-Differences Assumptions

We conducted several analyses to evaluate the validity of the difference-in-differences (DiD) design. First, DiD assumes that the trends in the treated and control groups over time would have remained similar in the absence of the EITC expansion (known as the “parallel trends assumption”). While this counterfactual is not empirically testable, we checked whether the pre-expansion trends in the treated and control groups were parallel; non-parallel trends would indicate that the control group was not sufficiently similar to the treated group. As a further check of the DiD design, we tested for differential compositional changes between the treated versus the control group in expansion versus pre-expansion years.

Statistical Analysis

We estimated two separate types of quasi-experimental models. In the first model, we used difference-in-differences (DiD) analysis, a method well suited for evaluating policy impacts while accounting for underlying trends among a control group.^42,43 Analogous to an experimental design, the model compared pre-post trends in health among young adults ages 18–24 years (treated group) to pre-post trends among adults ages 25–34 years (control group). While the 2021 EITC policy expansion was targeted at young adults making below $25,000, we restricted our main specification to those making below $50,000 because income may be reported differently in BRFSS compared with tax forms.

DiD analysis involves regressing each outcome on an interaction term between a binary term for whether the individual was aged 18–24 years and a binary term for whether they were observed in BRFSS survey year 2022 (tax year 2021), while adjusting for the main terms and the covariates listed above. This main analysis compared the expansion in tax year 2021 (BRFSS survey year 2022) with both the pre-expansion (tax year 2020, survey year 2021) and post-expiration tax years (tax year 2022, survey year 2023). However, we also separately examined the impacts of only the expansion in tax year 2021 (compared with tax year 2020) and only the expiration in tax year 2022 (compared with tax year 2021). The expansion and expiration occurred in different years with differing policy climates, and turning on a policy may have different impacts than turning off a policy.

In the second model, we implemented an event study analysis, which examines time in a more granular fashion relative to a standard DiD model. In other words, DiD gives overall effects in the expansion and the non-expansion periods, whereas the event study provides estimates for each month using a set of leads and lags.

The equation for each analysis and additional details are provided in the Appendix. Analyses were conducted using Stata/MP 18.5. Estimates were considered statistically significant at p < 0.05.

Secondary Analyses

We conducted a series of falsification (i.e., placebo) tests to estimate the validity of our DiD design. In the first falsification test, we re-estimated the model only among those with incomes above $50,000, comparing adults ages 18–24 to those ages 25–34. Higher-income groups were not eligible for the expanded EITC, and we would therefore expect the results to be the same in tax year 2021 compared with other years (i.e., null findings). In the second falsification test, we re-estimated the model using adults aged 25–29 years with incomes below $25,000 as a “placebo” treated group (since they were already eligible for the EITC before the expansion), while those who are 30–39 years served as the control. We would expect null findings in this falsification test.

Additionally, we performed further analyses to check the robustness of our model. First, we re-estimated the model narrowing the age group of our control group to 25–29-year-olds. Second, we re-estimated the model by including those who are unemployed but excluding students because they could be claimed by their parents as dependents. Third, we conducted an additional sensitivity analysis omitting the states (Colorado, California, Maine, Maryland, Minnesota, New Jersey, New Mexico) and DC that offer an EITC to young childless adults. Fourth, policies may have heterogeneous treatment effects among diverse subgroups.⁴⁴ We therefore performed subgroup analyses by stratifying models by race/ethnicity and sex. Finally, we also examined different income cut-offs, including those with incomes below $25,000 and those with incomes below $35,000; this allowed for a closer approximation to the actual income eligibility criteria for the 2021 EITC expansion but limited our sample size.

Ethical Approval

The study did not require Institutional Review Board approval as data were de-identified and publicly available.

Results

Descriptive Statistics

Table 1 shows the sample descriptive statistics for childless adults ages 18–24 years and 25–34 years during tax year 2021 (policy expansion in place) versus tax years 2020 and 2022 (policy expansion not in place). Young adults aged 18–24 years were more likely to have lower education and income compared to those aged 25–34 years. During tax years 2020 and 2022, young adults aged 18–24 years had 8.2 poor mental health days on average, and 27.7% experienced frequent mental distress. Adults aged 25–34 years had 7.3 poor mental health days, and 23.8% experienced frequent mental distress. Meanwhile, 27.7% of adults aged 18–24 years engaged in binge drinking compared with 25.3% of adults aged 25–34 years. Importantly, DiD does not require that the treated and control groups be identical during the pre-expansion period; see below for tests of DiD model assumptions.

Table 1:

Sample Characteristics, by Age and Year

	Childless adults 18–24 years		Childless adults 25–34 years
	Tax years 2020, 2022	Tax year 2021 (Expansion)	Tax years 2020, 2022	Tax year 2021 (Expansion)
	Percent
OUTCOMES
Poor mental health days	8.2	8.4	7.3	7.6
Frequent mental distress	27.7	28.4	23.8	26.0
Poor physical health days	2.5	2.8	2.8	3.5
Binge drinking	27.7	27.3	25.3	31.5
COVARIATES
Race/ethnicity
Non-Hispanic White	55.5	55.3	51.2	52.3
Non-Hispanic Black	9.9	10.2	14.8	14.1
Non-Hispanic Other	9.9	12.2	11.0	11.9
Hispanic	24.7	22.3	23.0	21.7
Female	42.9	43.9	36.3	38.2
Married	9.1	6.7	15.1	15.4
Education
Less than high school	8.5	7.9	10.8	10.3
High school	47.2	46.9	32.1	29.8
Some college	29.7	33.0	30.8	32.7
College or higher	14.6	12.2	26.3	27.2
Income
Less than $10,000	4.1	5.1	2.5	1.9
$10,000– $14,999	4.5	2.5	3.9	3.4
$15,000– $19,999	6.6	6.5	6.4	6.7
$20,000– $24,999	14.4	12.3	11.2	12.8
$25,000– $34,999	34.5	31.7	31.2	30.4
$35,000– $49,999	35.9	41.9	44.8	44.8
Metro	85.3	82.5	86.1	83.1
Insurance	84.3	83.6	76.4	78.1
Number of Observations	3,079	1,591	4,361	2,106

Notes: Data were drawn from the 2021–2023 waves of the Behavioral Risk Factor Surveillance System (BRFSS), including adults aged 18–34 years old who were interviewed between February-June, and were employed at the time of the BRFSS survey and who earned income below $50,000.

DiD Assumptions

Event study models demonstrated parallel trends in the treated versus control groups during the pre-expansion period (Figures 1–3), with the pre-policy confidence intervals including zero. This provides reassurance of the validity of the parallel trends assumption, although there was a (statistically insignificant) trend for mental health. We did not find any differential compositional changes in the characteristics of the treated versus control group during the expansion versus pre-expansion years (Supplemental Table 1).

Figure 1: Event Study Estimates of the 2021 Expansion Earned Income Tax Credit on Poor Mental Health Days and Frequent Mental Distress.

Notes: This graph shows the event-study estimates (dots) and the 95% confidence intervals (whiskers) of the effects of 2021 Earned Income Tax Credit (EITC) expansion among young adults without children ages 18–24 years versus older adults 25–34 years. The sample includes adults aged 18–34 years old with income below $50,000. All regressions controlled for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro status, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days. BRFSS, Behavioral Risk Factor Surveillance System; EITC, earned income tax credit.

Figure 3: Event Study Estimates of the 2021 Expansion of Earned Income Tax Credit on Engaging in Binge Drinking.

Notes: This graph shows the event-study estimates (dots) and the 95% confidence intervals of the effects of childless Earned Income Tax Credit (EITC) in 2021 on poor binge drinking. The sample includes adults aged 18–34 years old with income below $50,000. All regressions control for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days. BRFSS, Behavioral Risk Factor Surveillance System; EITC, earned income tax credit.

Main Analysis

In the DiD analysis comparing the expansion year with the pre- and post-expansion years (Table 2), there were no effects of the 2021 childless EITC expansion on health and health behaviors among those making under $50,000 annually, nor in the analyses using the narrower samples of individuals making less than $25,000. When we separately examined the impacts of the expansion and its expiration (Table 3), we found that the expansion was associated with 1.08 fewer days in poor mental health (95%CI: −1.94, −0.23), and a 4.7 percentage-point decline in frequent mental distress (95%CI: −0.086, −0.0082). Meanwhile, we were unable to reject the null hypothesis of no effect of the expiration of the policy.

Table 2:

Health Effects of the 2021 Temporary Childless EITC Expansion, Versus Pre-Expansion and Post-Expiration Periods

		Income Cut-off for Analytic Sample
Outcomes	Policy Effects	<$25,000	<$35,000	<$50,000
Mental health	Coefficient (95% CI)	−1.07	−0.48	−0.29
		(−2.75, 0.62)	(−1.54, 0.58)	(−1.05, 0.48)
	Sample size	2,612	6,250	11, 137
Frequent mental distress	Coefficient (95% CI)	−0.013	−0.013	−0.014
		(−0.086, 0.061)	(−0.060, 0.034)	(−0.049, 0.021)
	Sample size	2,612	6,250	11,137
Physical health	Coefficient (95% CI)	−0.31	0.051	0.030
		(−1.47, 0.84)	(−0.63, 0.74)	(−0.46, 0.52)
	Sample size	2,619	6,263	11,156
Binge drinking	Coefficient (95% CI)	−0.0078	−0.012	−0.021
		(−0.083, 0.067)	(−0.061, 0.037)	(−0.058, 0.016)
	Sample size	2,462	5,929	10,592

Notes: Effects are estimated from difference-in-differences models and represent the pre-post differences in each outcome in the treated group (ages 18–24 years) relative to the control group (ages 25–34 years). All regressions controlled for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro status, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days.

Table 3:

Health Effects of the 2021 Temporary Childless EITC Expansion, Versus Pre-Expansion and Post-Expiration Periods Separately

Outcomes	Policy Effects	Tax year 2020 (Reference, coded 0) vs. 2021	Tax year 2021 (Reference, coded 0) vs. 2022
Mental health	Coefficient (95% CI)	−1.08*	−0.66
		(−1.94, −0.23)	(−1.57, 0.25)
	Sample size	7,673	7,161
Frequent mental distress	Coefficient (95% CI)	−0.047*	−0.026
		(−0.086, −0.0082)	(−0.067, 0.016)
	Sample size	7,673	7,161
Physical health	Coefficient (95% CI)	−0.22	−0.35
		(−0.75, 0.32)	(−0.96, 0.26)
	Sample size	7,698	7,166
Binge drinking	Coefficient (95% CI)	−0.027	0.013
		(−0.069, 0.016)	(−0.031, 0.056)
	Sample size	7,261	6,803

Note: The effects are estimated from Difference-in-Differences model from Equation (1). The sample includes childless adults aged 34 years or less who were employed at the time of the BRFSS survey and earned below $50,000. All regressions controlled for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro status, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days.

^*p<0.05

The event study analyses confirmed reductions in the number of poor mental health days and frequent mental distress in several months during the expansion period (Figure 1). There was also a short-term reduction in poor physical health days showing declines in poor physical health days in April 2022, which reverted to pre-policy levels in the subsequent months (Figure 2). There was no change in binge drinking (Figure 3). These models also suggested that differences in mental and physical health between the treated and control groups did not fully return to baseline in tax year 2022 (survey year 2023). Supplemental Figures 2a–b, 3 and 4 show the event study analyses directly comparing the policy expansion with the expiration, again suggesting no change in health and behavioral outcomes.

Figure 2: Event Study Estimates of the 2021 Expansion of Earned Income Tax Credit on Poor Physical Health Days.

Notes: This graph shows the event-study estimates (dots) and the 95% confidence intervals of the effects of childless Earned Income Tax Credit (EITC) in 2021 on poor physical health days. The sample includes adults aged 18–34 years old with income below $50,000. All regressions control for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days. BRFSS, Behavioral Risk Factor Surveillance System; EITC, earned income tax credit.

Secondary Analyses

Results from the placebo tests among those with incomes above $50,000 who were unlikely to be affected by the EITC expansion (Supplemental Table 2), coefficients were reassuringly smaller and statistically insignificant. The results from the falsification test using 25–29 years old as placebo treated group compared to control group who were 30–39 years old showed null results (supplemental table 3) pointing to the validity of our DiD design. In subgroup analyses by sex (Table 4), there was a reduction in the number of poor mental health days and frequent mental distress among males and decreased binge drinking among females. In subgroup analyses by race/ethnicity, there were no effects among any individual subgroup. Results from our additional analysis using 25–29 year-olds as the control group (supplemental table 4) showed similar results to our main estimates, and we also observed a reduction in binge drinking. The results from our model including those who are unemployed but excluding students showed virtually identical results to our main estimates (Supplemental Table 5). Finally, the results for the sensitivity analysis omitting states that already offered EITC to childless young adults, and the analysis in which we did not adjust for income were similar to our main analysis (Supplemental Table 6 and 7).

Table 4:

Health Effects of the 2021 Temporary Childless EITC Expansion, Subgroup Analysis by Sex and Race/Ethnicity

	Coefficient (95% CI)
Outcomes	Sex		Race/ethnicity
	Female	Male	Non-Hispanic White	Non-Hispanic Black	Hispanic	Non-Hispanic Other
Poor mental health days	−0.87	−1.22*	−1.05	−0.57	−1.20	−1.28
	(−2.26, 0.52)	(−2.31, −0.13)	(−2.12, −0.025)	(−3.81, 2.68)	(−3.22, 0.83)	(−4.04, 1.47)
Sample size	3,197	4,476	4,978	645	1,200	850
Frequent mental distress	−0.035	−0.056*	−0.034	−0.037	−0.049	−0.12
	(−0.10, 0.031)	(−0.10, 0.008)	(−0.084, 0.016)	(−0.18, 0.10)	(−0.14, 0.044)	(−0.24, 0.003)
Sample size	3,197	4,476	4,978	645	1,200	850
Poor physical health days	0.093	−0.39	−0.001	0.53	−1.18	−0.76
	(−0.77, 0.95)	(−1.07, 0.28)	(−0.67, 0.67)	(−1.43, 2.49)	(−2.44, 0.076)	(−2.43, 0.91)
Sample size	3,201	4,497	5,005	646	1,198	849
Binge drinking	−0.072*	0.004	−0.027	−0.060	−0.024	−0.038
	(−0.14, −0.007)	(−0.052, 0.060)	(−0.080, 0.027)	(−0.20, 0.081)	(−0.13, 0.085)	(−0.17, 0.092)
Sample size	3,019	4,242	4,758	593	1,109	801

Notes: The effects are estimated from Difference-in-Differences model from Equation (1). The sample includes childless adults aged 34 years or less who were employed at the time of the BRFSS survey and earned below $50,000. All regressions controlled for age, sex, race/ethnicity, education, marital status, household income, insurance status, metro/non-metro status, time fixed effects, and state fixed effects. Sample was drawn from the Behavioral Risk Factor Surveillance System (BRFSS, N = 11,137). Frequent mental distress represents those with 14 or more poor mental health days in the last 30 days.

^*p<0.05, indicating that an estimate was statistically significantly different from zero.

Discussion

Our study provides some of the first evidence of the health impacts of the 2021 temporary EITC expansion among young adults without children. We found the expansion led to decreased mental distress in the overall sample when the policy was implemented. We did not see inverse effects after the expansion expired, perhaps due to co-occurring policy changes such as the end of the public health emergency and termination of pandemic benefits for many groups in early 2023 that also affected individuals in the control group, thereby reducing our ability to detect effects in the treated group. We also found reductions in poor mental health among males, although these were not statistically significantly different from results among females.

Only two prior studies have examined the impact of the 2021 childless EITC expansion, finding a reduction in housing hardship, and improvement in mental health among young adult renters.^33,34 Reduced material hardship may be a mechanism through which mental health improved in our study. At the same time, the more limited impact on other health outcomes maybe because of the modest size of the 2021 EITC expansion relative to the greater economic hardship brought about by the pandemic. Alternately, our sample may not have been large enough to detect small effects. It may also be that take-up of the 2021 EITC expansion was more limited among young adults because of their inexperience with the tax filing process, thereby limiting the EITC’s positive impacts; prior work has found that EITC take-up in general is lower among young adults.⁴⁵ Also, although the control group of 25–34-year-olds was EITC-eligible prior to the 2021 tax year, they also experienced an increase in EITC benefits during 2021, which may have led to a less stark difference in the exposure between the treated and control groups, resulting in muted impacts. Similar explanations have been hypothesized for the null results observed in the Paycheck Plus Atlanta Study.³²

Other research has found that policy expansions during the pandemic—such as the CTC expansion and Supplemental Nutrition Assistance Program emergency allotments—improved mental health, although those policies are more likely to affect individuals with children.^46–48 Other studies showed mixed results of the CTC expansion, with limited impact on mental and physical health of adults.⁴⁹ Meanwhile, prior research on the EITC has also generally found improved mental health and health behaviors, but those studies did not include young adults without children who were previously ineligible for EITC benefits.^24,25 Even before the pandemic, younger adults experienced greater socioeconomic disparities and greater risk for mental health problems and substance use.^50–52 Our study provides important evidence on a key policy to address health among this vulnerable group at a critical point in the life course.

The stratified analyses demonstrated larger effects among males, but racial/ethnic subgroup analyses were null. This may be because of limited power to detect modest effects among these smaller subgroups. Our findings contrast with prior work, which has shown differences in EITC impacts by race/ethnicity, sex, and education.^53–56 Future studies should examine this research question in larger samples.

Our study had some limitations. First, BRFSS age categories include those aged 18 to 24 years, which is slightly different than the EITC eligibility criteria of 19 to 24 years; therefore, there is a small degree of measurement error in classifying individuals exposed to the policy expansion. Second, income, health, and covariates are self-reported in BRFSS and may suffer from measurement error and standard reporting biases. Third, because of the dynamic policy landscape during the COVID-19 pandemic, there may be confounding due to other co-occurring policies or events.⁵⁷ We address this limitation by conducting several assumption tests and secondary analyses (which were reassuring), but this is a limitation of all DiD studies. Also, while parallel trends were reassuring, there was a (statistically insignificant) trend for mental health during the pre-period in event study analyses, so results should be interpreted with caution. Finally, BRFSS data lacks individual-level controls such as utilization of health care services including mental health services. But we included in our model educational attainment and having insurance coverage, variables that are correlated with income.

In conclusion, our study evaluated the impact of the 2021 EITC expansion on health and health behaviors of young adults without children. We found evidence that the EITC expansion mainly improved mental health with little improvement in physical health, and limited differential impacts among subgroups. The modest improvement in physical health could be attributed to the temporary nature of the policy providing brief income support that may not have been enough to change the physical health of young adults, who are already relatively healthy. The findings point to potential health benefits of this novel (but temporary) income support policy targeted at vulnerable young adults who face considerable hardships and yet have limited social safety net supports compared to other age groups. This topic remains actively discussed at the federal and state levels; e.g., Maine expanded its EITC program to childless young adults in 2019.⁵⁸ Young adults who are ineligible for many safety net programs would likely benefit from EITC expansions that reduce poverty as a key social determinant of health.