Association between Child Tax Credit advance payments and food insufficiency in households experiencing economic shocks

Abstract The COVID-19 pandemic brought increases in economic shocks due to poor health and lost employment, which reduced economic well-being, especially in households with children. The American Rescue Plan Act of 2021 expanded Child Tax Credit (CTC) payments to include eligibility for the lowest income households, boosted benefit levels, and provided monthly advance payments to households with children. Using Census Household Pulse Survey respondent data from January 2021 to July 2022, we evaluated the association between these advance CTC monthly payments and food insufficiency among households with children experiencing health- or employment-related economic shocks (defined as missed work due to COVID-19/other illness or COVID-19–related employer closure/layoff/furlough). Using a triple difference design, we found that the advance CTC was associated with greater reductions in food insufficiency among households with children experiencing economic shocks both compared with households without children and with households with children not experiencing economic shocks. Permanently expanding the advance CTC could create resilience to economic shocks during disease outbreaks, climate disasters, and recessions.


Introduction
2][13][14] Thus, policies that can address food insufficiency among populations vulnerable to economic shocks are key for achieving health equity.
The COVID-19 pandemic brought widespread economic shocks across the United States, with more than 50 million people losing employment.Additionally, missing work due to illness increased by 50% compared with the 2 years prior. 15,16Negative effects were disproportionately concentrated among low-income, Black, Hispanic, and/or families with children who were most likely to miss work and less likely to receive paid sick leave. 15Further, these groups were also more likely to experience increases in food insufficiency. 12,13o mitigate negative economic consequences of COVID-19, Congress passed the American Rescue Plan Act (ARPA) in March 2021.The ARPA included 3 modifications to the existing Child Tax Credit (CTC): (1) eligibility for the full CTC amount was expanded to families with low/no income; (2) the maximum credit amount increased from $2000 to $3000 per qualifying child for children aged 6-17 years (previously ended at 16 years) and to $3600 for children aged 5 years or younger, with greater gains for low-income households; and (3) payments were delivered as monthly per-child advances of $200-$300 during the 6-month period from July 2021 through December 2021.Families received the rest of the increased CTC amount for 2021 when filing taxes in 2022.In early 2022, the CTC expansion expired and reverted to its original structure, with lower credit amounts, no monthly advance payments, and non-refundability, excluding those with the lowest incomes.
Families reported using advance CTC monthly payments for necessities such as food and clothing, and payments did not lead to decreased labor supply among recipients. 17,180][21][22] However, it is still unclear if the advance CTC had differential impacts for vulnerable households, such as those experiencing economic shocks.
The objective of our study was to investigate how the ARPA expansion of the CTC, particularly the advance payments, impacted food insufficiency, including how it may have differentially protected households vulnerable to COVID-19-driven economic shocks.We aim to expand on prior literature to understand how the advance CTC may differentially affect vulnerable households and why it may have differential effects across demographic groups, including how the payments may support households to maintain food sufficiency and increase resilience to economic shocks in periods such as disease outbreaks, climate disasters, or recessions.

Sample and study period
We used the US Census Bureau's nationally representative Household Pulse Survey (HPS), which contains respondent self-reported demographics, household composition, employment, and economic hardship data (including food insufficiency and participation in various assistance programs). 23n HPS, only 1 adult responds on behalf of the household.We limited the sample to adults younger than 65 years to capture those of working age with primarily their own children in the household, and to individuals without missing data (except for income, for which a missing category was created). 19,20We included data from HPS waves 22-27, capturing 3 time periods: ( 1

Exposure: advance CTC payments
Through the ARPA in 2021, over 90% of families with children were eligible to receive $200-$300 monthly per child between July and December 2021.Families who received monthly payments were eligible to claim an additional $1500-$1800 per child after filing their 2021 tax return in early 2022. 19,24,25he advance CTC payment exposure was defined as living in a household with at least 1 child present during the survey period that covered months when the advance CTC payments were being dispersed (July 21, 2021-January 10, 2022).

Exposure: economic shocks
We defined economic shocks as a report of at least 1 healthrelated shock or employment-related shock.Health-related shocks were defined as a report of missing work in the past 7 days for 1 of the following reasons: "I am/was sick with coronavirus symptoms or caring for someone sick with coronavirus symptoms" or "I am/was sick (not coronavirus-related) or disabled."Employment-related shocks were defined as a report of missing work in the past 7 days for 1 of the following reasons: "I am/was laid off or furloughed due to the coronavirus pandemic," or "my employer closed temporarily due to the coronavirus pandemic," or "my employer went out of business due to the coronavirus pandemic."

Outcome
The outcome was household food insufficiency, defined as a binary measure.Following US Department of Agriculture (USDA) standard methods for coding survey responses, 26 food insufficiency was defined as a report of "sometimes not enough to eat" or "often not enough to eat" in response to the following question: "In the last seven days, which of these statements best describes the food eaten in your household?"Respondents were not considered to have food insufficiency if they reported "enough of the kinds of food I/we wanted to eat" or "enough, but not always the kinds of food (I/we) wanted to eat" in response to this question. 23,26he HPS food-insufficiency measure of not having enough food to eat in the past 7 days is related to food insecurity, which is more expansive and based on a scale developed by the USDA.Food insufficiency is a narrower definition that focuses on food-intake quantity. 26Researchers used HPS data on food insufficiency to provide real-time information throughout the pandemic; several studies have used the HPS measure to gain insight into food access during the COVID-19 pandemic. 12,19,20,27,28variates Other relevant policy changes during the study period included the third and final Economic Impact Payment (EIP) in March 2021, the cessation of unemployment insurance expansions in June-September 2021, and the expiration of the federal eviction moratorium in August 2021.We adjusted for self-reported individual receipt of the EIP and/or unemployment insurance and eviction risk (somewhat/very likely) to account for these changes.We included covariates for participation in other public assistance benefits, including the Supplemental Nutrition Assistance Program (SNAP) and other food aid based on yes/ no questions, as well as respondent health insurance coverage.We further adjusted for demographic characteristics, including sex assigned at birth, age group, educational level, prior-year household income, marital status, number of adults in the household, and number of children in the household.We also included survey wave and state fixed effects, which capture national trends that affected both households with children and households without children during each time period (eg, inflation) as well as time-invariant state characteristics not captured by other covariates.

Analyses
We reported the unadjusted prevalence of any economic shocks, health-related shocks, and employment-related shocks prior to advance CTC implementation for all households, households with and without children, and stratified by subgroup (by race/ethnicity and for low-income households [earning <$35 000 in the prior year 20,21 ]).
We also reported the unadjusted prevalence of household food insufficiency before, during, and after advance CTC implementation for (1) all households, (2) households without children not experiencing economic shocks, (3) households with children not experiencing economic shocks, (4) households without children experiencing economic shocks, and (5) households with children experiencing economic shocks, and by racial/ethnic and income subgroup.
We conducted linear regressions to evaluate the association between experiencing any economic shocks, health-related shocks, and employment-related shocks and household food insufficiency in the period prior to advance CTC implementation for all households, households with and without children, and by racial/ethnic and income subgroup.We adjusted for wave and state fixed effects, time-varying covariates including receipt of EIP, SNAP, and demographic characteristics (detailed above).
In our primary model, we conducted a difference-indifference-in-differences ("triple difference") analysis of household food insufficiency.We compared changes in household food insufficiency in the period during advance CTC implementation in households with children experiencing economic shocks relative to households without children and to households not experiencing economic shocks.The main exposure was a binary indicator for being in a household with children, interacted with a binary indicator for the period during advance CTC implementation, and with a binary indicator for households experiencing economic shocks.The resulting estimates speak to the association between the advance CTC and food insufficiency in households with children compared with those without children, and if this association varied depending on whether households were experiencing economic shocks.We used linear models and adjusted for state and wave fixed effects, demographics, and time-varying covariates described above (Appendix Methods).Triple difference relies on the parallel trends assumption, which assumes that underlying trends between households with and without children would be parallel in the absence of the advance CTC.To test this assumption, we evaluated whether differences in food-insufficiency trends varied over time between households with and without children in the period prior to advance CTC implementation.We tested this association using both a triple difference estimate (ie, interacting an indicator for the presence of children with an indicator for the presence of economic shocks with a continuous time indicator) and difference-in-difference estimates separately among households (1) experiencing and (2) not experiencing economic shocks (ie, interacting an indicator for the presence of children with a continuous time indicator in both models).We found that, before advance CTC implementation, there were no significant differences in food-insufficiency trends over time in naive and adjusted versions of all 3 models (Table S1).
To test the robustness of our findings, we clustered standard errors by state and varied inclusion of fixed effects.We also conducted 2 additional difference-in-difference analyses.We evaluated the interaction between advance CTC implementation and children in the household by estimating stratified models separately among households experiencing and not experiencing shocks.Additionally, some respondents may have received lump-sum tax refunds during the period after the advance CTC monthly payment expiration, depending on when they filed their tax return.Therefore, we conducted a sensitivity analysis in which we excluded dates after the tax filing deadline in April 2022 based on evidence that most Americans file their taxes close to the deadline. 29Last, we conducted the triple differences analysis limited to the low-income subgroup only, because this group was most likely to receive the greatest benefit from changes to the CTC made in the ARPA. 21,22his study was exempt from Institutional Review Board (IRB) approval and informed consent and meets Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for crosssectional studies. 30In all models, we used US Census-provided weights divided by the number of survey waves.We used 2-sided t tests or chi-square tests to test for significant differences between groups; P < .05 was considered significant.Analyses were conducted April through October 2023 using Stata/MP version 17.0 (StataCorp, College Station, TX).

Sample characteristics
Our sample comprised 1 125 299 respondents, representing a weighted population of 122 800 808 individuals.Respondents were majority female (n = 691 775 [51.3%]) and non-Hispanic White (n = 808 207 [62.2%]), with a plurality (n = 463 735 [48.4%]) aged 25-44 years.Weighted individual demographics and household socioeconomic characteristics are shown in Table 1 for the full sample and stratified by household presence of children and experience of economic shocks.

Unadjusted household food insufficiency prevalence before, during, and after advance CTC implementation
As depicted in Figure 1, unadjusted household foodinsufficiency prevalence was elevated among households experiencing economic shocks (solid lines) relative to households not experiencing economic shocks (dashed lines).In both circumstances, food insufficiency was higher in households with children (green lines) relative to households without children (gray lines).While food insufficiency increased over the study period, there was a temporary dip in food insufficiency in households with children during the period of advance CTC implementation.Overall, Hispanic, non-Hispanic Black, and low-income subgroups had the highest unadjusted household food insufficiency.Household food-insufficiency prevalence by race/ethnicity and low-income subgroups is shown in Figure 2 and Table S3.Except for Asian and "another racial category" respondents, the presence of children in the household was associated with similar or higher rebound rates of food insufficiency post-advance CTC expiration.

Difference-in-difference analyses: estimated association between advance CTC implementation and food insufficiency among households with children experiencing economic shocks
Our triple difference model estimated a 3.5 percentage point decrease (95% CI: 6.1-0.90 percentage points; P = .008)in the proportion of respondents who reported experiencing food insufficiency among households with children experiencing economic shocks compared with households without children and with those not experiencing shocks during the period of advance CTC implementation (Table 2, Table S5).This represents an 11% decrease compared with the period before advance CTC implementation.Results were robust to clustered standard errors and varied inclusion of fixed effects (Table S6).
Additionally, difference-in-difference models conducted separately among households experiencing and not experiencing  Health Affairs Scholar, 2024, 2(2), 1-9 economic shocks were consistent with triple difference results.
In both groups, the advance CTC was associated with decreased household food insufficiency, but the decrease was larger for households experiencing economic shocks compared with households not experiencing shocks (5.2 and 1.6 percentage point decrease, respectively) (Table S7).
Our results held when we excluded the period after the 2022 tax filing deadline from the triple difference model, although the effect was slightly greater.When this period was excluded, there was a 3.9 percentage point decrease in the proportion of respondents reporting food insufficiency among households with children experiencing economic shocks compared with households without children and those not experiencing economic shocks (compared with a 3.5 percentage point decrease in the main analysis) (Table S8).When the triple difference analysis was limited to the low-income subgroup, the magnitude of the effect remained the same (a 3.5 percentage point decrease), but the effect was no longer significant, likely due to a large reduction (84% decrease) in sample size and resulting lack of precision (Table S9).

Discussion
We found that the advance CTC was associated with a 3.5 percentage point and 11% differential decrease in the probability of experiencing food insufficiency for households with children experiencing economic shocks compared with households without children and households not experiencing economic shocks.0][21] We additionally found that the advance CTC was associated with a greater reduction in food insufficiency among households with children experiencing economic shocks (ie, missing work due to sickness with COVID-19/other illness or employer closures/layoffs/ furloughs) than among those not experiencing economic    S3.Source: Authors' analysis of Household Pulse Survey data from the Census Bureau, January 2021 to July 2022.

6
Health Affairs Scholar, 2024, 2(2), 1-9 shocks.These results help to understand the mechanisms through which the advance CTC can protect against food insufficiency, through differentially impacting demographic groups most structurally vulnerable to economic shocks, including low-income, Hispanic, and Black populations.Such findings are relevant for federal and state policymakers, as Congress could consider restoration of the advance CTC and as state lawmakers consider similar models. 31,32These findings may also be relevant to broadly reducing vulnerability to food insufficiency in households with children, especially in preparation for events such as disease outbreaks, climate disasters, or recessions, which could produce economic shocks.
Our results parallel prior observational and quasiexperimental studies that have found that, both before and during the COVID-19 pandemic, unexpected events impacting employment contribute to economic hardship. 1For example, 1 study found a positive association between parental decline in employment and household food insecurity. 2 Additional state and national survey data indicate that people who lost or missed work due to COVID-19 were most likely to report worse food access or greater food insufficiency. 3,33,34owever, to our knowledge, our study is the first to evaluate how the relationship between the advance CTC and household food insufficiency varied by contemporaneous economic shocks experienced by household members.
Even before the COVID-19 pandemic, the United States had a higher rate of children living in poverty than most other Organization for Economic Cooperation and Development (OECD) countries, with disparities by race and ethnicity. 11,35][38] People with low income and wealth are vulnerable to food insufficiency if they face sudden health or employment shocks. 39his vulnerability had a particularly negative impact during the COVID-19 pandemic, when absences from work due to illness, child care, or other personal obligations increased by 50% compared with the 2 years prior, and low-income, non-Hispanic Black, and Hispanic employees were least likely to have paid sick leave. 15nsistent with prior findings, we observed similar socioeconomic and racial/ethnic disparities: during the COVID-19 pandemic, low-income, Black, and Hispanic populations disproportionately experienced economic shocks.In the period before advance CTC implementation, low-income, non-Hispanic Black, and Hispanic households were 151%, 68%, and 55% more likely to experience economic shocks compared with higher-income and White households, respectively.Similarly, we found that the prevalence of household food insufficiency was higher among low-income, non-Hispanic Black, and Hispanic households compared with higher-income White households over the study period.Our results suggest that these findings are related; experiencing economic shocks was associated with an 80% increase in household food insufficiency.In combination, our findings show that the advance CTC was associated with an especially protective impact on households experiencing economic shocks, which were disproportionately non-Hispanic Black, Hispanic, and low-income, suggesting that the advance CTC may be impactful in these racial, ethnic, and income groups.While prior work has described differential impacts of the advance CTC on food insufficiency across demographic groups, [20][21][22] our findings provide insight into how differential effects may arise: through protecting structurally vulnerable households (including those most likely to have low wealth/income and unpaid sick leave) in times of missed work due to sickness or job loss.
Our study has limitations.First, the HPS relies on selfreported, repeated cross-sectional data, with low overall response rates.Because the data are repeated cross-sectional surveys rather than longitudinal, we could only evaluate contemporaneous associations between economic shocks and food insufficiency.Second, we defined economic shocks based on available survey response options and could not capture individuals who experienced economic shocks for other reasons.Third, respondents may have received lump-sum tax refunds during the period after the advance CTC monthly payment expiration.In fact, we found that, when we excluded dates after the tax filing deadline from analysis, the effect size was slightly greater, suggesting that we may be underestimating the impact of the monthly payments on food insufficiency in the absence of lump-sum returns.However, the difference was not large, which may be consistent with prior evidence that lump-sum and monthly payments may be used for different purposes (lumpsum payments more often used for arrears and monthly payments for ongoing costs). 40Fourth, we estimated the effect of the advance CTC on households eligible for the advance CTC (ie, households with children), rather than households that received the advance CTC.We used this "intent-to-treat" approach to avoid selection bias, as households that filed taxes during the advance CTC may differ from those that do not.This is a conservative approach that may bias results towards the null; the effect size may be greater only among households that received the advance CTC.Another limitation is that there may have been time-varying differences between households with and without children, such as school closures, which could have impacted food insufficiency.Finally, our triple difference analysis was underpowered when restricted to low-income households only (a subgroup highly impacted by the ARPA changes to the CTC); however, while we did not observe significant effects in this subgroup, the magnitude of the effect size was the same as in our main analysis, despite the loss of precision.Because of the proxy identification of benefits from the advance CTC (households with children), we used non-causal language to be conservative, despite the use of a causally interpretable triple difference design.These limitations are counterbalanced by several strengths.First, we utilized data from a nationally representative, highfrequency survey, capturing variation and granularity in individuals' and households' circumstances.Second, we used a quasi-experimental triple difference approach, allowing us to build upon our and others' earlier work by exploring variation by contemporaneous economic shocks in addition to the presence of children in the household.

Conclusion
In conclusion, our study aligns with prior work demonstrating anti-poverty benefits of the ARPA's advance CTC, which simultaneously expanded eligibility to families with no or very low income, increased maximum benefit amounts, and provided monthly in addition to annual payments.We add to this body of evidence by showing that health-and employment-related economic shocks are associated with increased household food insufficiency and that the advance CTC was associated with decreased food insufficiency, particularly among households experiencing economic shocks.Because we found a higher prevalence of both economic shocks and food insufficiency among low-income, non-Hispanic Black, and Hispanic households, permanently reinstating the advance CTC could progress equity for households that are structurally vulnerable to economic shocks by reducing household food insufficiency and related negative long-term health and education consequences for children.Members of Congress may consider the protective effect that reinstating the advance CTC could have for US households in both the shortand long-term.Legislators may also consider that reintroducing the advance CTC could increase population resilience and better protect households with children from hardship caused by health, environmental, or economic circumstances.

Figure 1 .
Figure 1.Unadjusted prevalence of household food insufficiency among households with and without children experiencing and not experiencing economic shocks before, during, and after advance Child Tax Credit (CTC) implementation.The figure shows the unadjusted prevalence of household food insufficiency in the periods before advance CTC implementation (January 6, 2021, through July 5, 2021), during advance CTC implementation (July 21, 2021, through January 10, 2022), and after advance CTC implementation (January 26, 2022, through July 11, 2022).The solid green line represents households with children experiencing economic shocks (health-related or employment-related shocks).The solid gray line represents households without children experiencing economic shocks.The green and gray dotted lines represent households with and without children, respectively, not experiencing economic shocks.Prevalence values are shown above or below each line.The dip in food insufficiency during the period of advance CTC implementation seen among households with children experiencing economic shocks is indicated with an arrow.Source: Authors' analysis of Household Pulse Survey data from the US Census Bureau, January 2021 to July 2022.

Figure 2 .
Figure 2. Unadjusted prevalence of household food insufficiency among households with and without children experiencing and not experiencing economic shocks before, during, and after advance Child Tax Credit (CTC) implementation: stratified by race, ethnicity, and low-income subgroups.The figure shows the unadjusted prevalence of food insufficiency in the periods before advance CTC implementation (January 6, 2021, through July 5, 2021), during advance CTC implementation (July 21, 2021, through January 10, 2022), and after advance CTC implementation (January 26, 2022, through July 11, 2022), stratified by race, ethnicity, and low-income subgroups.The solid green lines represent households with children experiencing economic shocks (health-related or employment-related shocks).The solid gray lines represent households without children experiencing economic shocks.The green and gray dotted lines represent households with and without children, respectively, not experiencing economic shocks.Dips in food insufficiency during the period of advance CTC implementation seen among households with children experiencing economic shocks are indicated with an arrow.A full table of unadjusted household food-insufficiency prevalence values is available in TableS3.Source: Authors' analysis of Household Pulse Survey data from the Census Bureau, January 2021 to July 2022.

Table 1 .
Sample characteristics, overall and for households experiencing or not experiencing economic shocks, and with or without children.

Table 1 .
Continued Data are presented as n (weighted %).Values may not add up to 100% due to rounding.Source: Authors' analysis of Household Pulse Survey data from the Census Bureau, January 2021 to July 2022.a P values within the "households not experiencing shocks" and "households experiencing shocks" columns represent the difference between households without and with children within each category, respectively.values on the far-right column represent the difference between households experiencing and not experiencing shocks.
b P

Table 2 .
Change in household food insufficiency during implementation of advance CTC in households with and without children experiencing and not experiencing economic shocks.In triple difference analysis, comparison is between households with children and without children and between households experiencing shocks and not experiencing shocks. a