Effects of the 2021 Expanded Child Tax Credit on Parents’ Well-Being and Time Use — Laetitia Lebihan (2025)

Summary

A paper by Laetitia Lebihan, Effects of the 2021 expanded Child Tax Credit on parents' well-being and time use, published in Economics and Human Biology, 2025, 59, and deposited in the HAL archive on 13 Feb 2026. The abstract states that the American Rescue Plan Act of 2021 temporarily provided unconditional monthly cash benefits to most households with children, and that the author uses the American Time Use Survey and Well-Being Supplement to examine the effects of the Child Tax Credit expansion. It states that the expansion was associated with improved parental well-being and health and with a significant increase in the time the parent spends with the child.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Issuer: Source documents
Document type: Effects Of The 2021 Expanded Child Tax Credit On Parents Well Being And Time Use
Case: Effects Of The 2021 Expanded Child Tax Credit On Parents Well Being And Time Use

Cite this document

“Effects of the 2021 Expanded Child Tax Credit on Parents’ Well-Being and Time Use — Laetitia Lebihan (2025).” Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/effects-of-the-2021-expanded-child-tax-credit-on-parents-well-being-and-time-use/, SHA-256 0f5bf7e6b73a19db85a989cbb3ce0a86ec6787cc1554241a6d640b2828df58c8.

Full text

Effects of the 2021 expanded Child Tax Credit on parents’
well-being and time use
Laetitia Lebihan

To cite this version:
Laetitia Lebihan. Effects of the 2021 expanded Child Tax Credit on parents’ well-being and time use. Eco-
nomics and Human Biology, 2025, 59, pp.101533. ⟨10.1016/j.ehb.2025.101533⟩. ⟨hal-05509374⟩

HAL Id: hal-05509374
https://hal.science/hal-05509374v1
Submitted on 13 Feb 2026

HAL is a multi-disciplinary open access archive L’archive ouverte pluridisciplinaire HAL, est des-
for the deposit and dissemination of scientific re- tinée au dépôt et à la diffusion de documents scien-
search documents, whether they are published or not. tifiques de niveau recherche, publiés ou non, émanant
The documents may come from teaching and research des établissements d’enseignement et de recherche
institutions in France or abroad, or from public or pri- français ou étrangers, des laboratoires publics ou
vate research centers. privés.

Distributed under a Creative Commons CC BY 4.0 - Attribution - International License
Economics and Human Biology 59 (2025) 101533

Contents lists available at ScienceDirect

Economics and Human Biology
journal homepage: www.elsevier.com/locate/ehb

Full length article

Effects of the 2021 expanded Child Tax Credit on parents’ well-being and
time useI
Laetitia Lebihan
University of Reunion Island, Department of Economics (CEMOI), 15 avenue René Cassin - CS 92003. 97744 Saint-Denis Cedex 9, France

ARTICLE INFO ABSTRACT

JEL classification: The American Rescue Plan Act of 2021 temporarily provided unconditional monthly cash benefits to most
H51 households with children to reduce child poverty during the COVID-19 pandemic. Using the American Time Use
I18 Survey and Well-Being Supplement, we examine the effects of the 2021 Child Tax Credit (CTC) expansion on
J18
well-being and time-use activities of households with children. We find that the CTC expansion was associated
Keywords: with improved parental well-being and health. The analysis also shows a significant increase in the time the
Child tax credit parent spends with the child. The results are robust to several robustness checks and consistent with existing
Well-being
evidence.
Health
Time use

1. Introduction symptoms (Pignatti and Parolin, 2024). However, the reform had no
significant effects on parents’ employment or labor force participa-
The COVID-19 pandemic has caused severe financial hardship for tion (Ananat et al., 2024; Pac and Berger, 2024).
the U.S. population due to sharp increases in unemployment rates and In this paper, we exploit data from the American Time Use Survey
declining incomes. Many studies have shown a significant deterioration (ATUS) and Well-Being (WB) Supplement module to estimate the im-
in people’s mental health and well-being during the pandemic (Ettman pact of the 2021 CTC reform on parents’ well-being and time use. We
et al., 2020; Vindegaard and Benros, 2020). In response to these chal- impute a CTC value to proxy how much different families would likely
lenges, the American Rescue Plan (ARP) Act temporarily expanded the benefit from the policy change and include this predicted CTC value in
federal Child Tax Credit (CTC) in 2021. The benefit size was increased a difference-in-differences (DD) model. Specifically, in the model, CTC
from $2000 per child to $3600 per child for children 0–5 years old and benefits vary by the number of children in the household, the children’s
to $3000 per child for children 6–17 years old. Moreover, the credit was ages, and the period studied.
made fully refundable, and eligibility was extended to households with This study contributes to the existing literature on child benefits in
little or no earnings. The 2021 CTC increased access to the benefits two main ways. First, to the best of our knowledge, this is the first
for approximately 26 million children whose families had previously study to provide an empirical analysis of the impact of CTC expansion
been partially or completely ineligible for the credit. The first monthly on well-being using ATUS and WB Supplement data. The ATUS data
payment went to more than 59.3 million children in July 2021, while
contain measures of daily activities. In the WB Supplement module, the
subsequent payments reached more than 61 million children, covering
survey randomly selects three activities reported by each respondent.
90% of American families with children (Department of Treasury,
For each of the selected activities, respondents were asked how they
2021).
felt along five dimensions (presented in a randomized order): happy,
Existing literature on the effects of CTC expansion has shown a de-
stress, pain, sad, and tiredness. Respondents were also asked questions
crease in child poverty and food insecurity among parents (Pilkauskas
about their general health. These data are widely used in the literature,
et al., 2024a). Families also spend more on food, housing, and child-
but have not been used to study the effect of the 2021 CTC reform.
related goods and services (Schild et al., 2023). The CTC expansion
Second, we also investigate the effects of the 2021 CTC reform on
improved parents’ mental health outcomes (Cha et al., 2023), includ-
ing fewer bad mental health days and fewer depressive and anxiety parental time use. To the best of our knowledge, this is the first study to

I This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. ATUS and WB Supplement data are
available from: https://www.bls.gov/tus/data-overview.htm. All computations on these microdata were prepared by the author, who assumes the responsibility
for the use and interpretation of these data.
E-mail address: laetitia.lebihan@univ-reunion.fr.

https://doi.org/10.1016/j.ehb.2025.101533
Received 16 March 2025; Received in revised form 5 September 2025; Accepted 9 September 2025
Available online 16 September 2025
1570-677X/© 2025 The Author. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
L. Lebihan Economics and Human Biology 59 (2025) 101533

provide an empirical analysis of the impact of unconditional monthly payments from July 2021 to December 2021 and the other half as a
cash transfers on time-use activities among the U.S. population. This lump-sum payment at tax time in spring 2022.
analysis is essential for developing a comprehensive evaluation of the The full credit was available to households with modified adjusted
reform and determining whether parents spend more time with and gross income (AGI) lower than $112,500 for single households and
invest more resources in their children. Thus, we contribute to a small $150,000 for married households for the 2021 tax year (including those
but growing collection of literature that seeks to investigate the effects with zero earned income). When income exceeded these thresholds, the
of universal child benefits on parental well-being and time use in credit was phased out. The first phase-out occurs when incomes exceed
North America. Contrary to most other cash assistance programs in the these thresholds but are below $200,000 for single filers and $400,000
U.S. (e.g., Earned Income Tax Credit), the 2021 reform temporarily for married filers. In this phase-out, the total benefit per child was
transformed the CTC into a near-universal unconditional cash trans- reduced by $50 for each $1000 additional AGI. The credit cannot fall
fer to families with children for 2021. Our paper offers important below $2000 (i.e., the same amount that eligible families received for
implications for policymakers and could contribute to the ongoing under the previous law). The second phase-out applied to filers with
debate regarding universal child benefits in the U.S. In addition, many AGI more than $200,000 and $400,000 for single and married filers,
studies have analyzed the association between extra family income and respectively. In this phase-out, the total benefit per child was reduced
well-being/health, but few have examined causality. Here, we exploit by $50 for each $1000 additional AGI. The credit could decrease below
exogenous policy-induced variations in family income to estimate the $2000 until reaching $0.
causal impact of income on several measures of well-being and time Fig. 1 shows the effect of the reform on the CTC benefit structure.
use. The figure compares the 2021 expanded CTC with the pre-reform
We find that the CTC expansion led to a significant decrease in CTC by household income and age of children for single and married
sadness for parents, but no significant changes in happiness, stress, households.
pain, or tiredness measures. Estimates suggest that the reform improved In order to receive their monthly payments, eligible recipients had
parents’ self-reported health. The analysis also shows a significant to have filed their taxes or registered for payments through the CTC
increase in the time parents spent with the child (called childcare in portal hosted by the Internal Revenue Service (IRS). If bank account
the paper later). Our findings are robust to several robustness checks information was known, the funds were transferred directly to eligible
and consistent with existing evidence on the relationship between CTC recipients by the IRS. Evidence shows that, between 2020 and 2021, the
expansion and well-being. child poverty rate decreased by about half (from 9.7% to 5.2%), with
The remainder of the paper is organized as follows. Section 2 de- 90% of this decrease being attributed to the CTC according to the U.S.
scribes the 2021 CTC expansion. Sections 3 and 4 describe, respectively, Census Bureau (Curran, 2022). The expanded CTC was implemented
the sample data and the methodology. In Section 5, we present the only for the 2021 tax year. Monthly benefits ended in December 2021,
empirical results. In Section 6, we discuss the results and conclude the and households claimed the remainder of the benefits when filing taxes
paper. in spring 2022. This program was not renewed by Congress, and the
CTC reverted to its original form as of the 2022 tax year.
2. Reform
3. Data
Established in 1997, the CTC was intended to provide financial
assistance to middle-income families.1 Over time, the amount of the We use data from the 2021 ATUS and WB. The ATUS data are an
credit increased and, as of 2017, tax filers could receive a maximum annual cross-sectional survey that involves a representative sample of
CTC of $2000 per child per year. However, many children did not the U.S. population and is run by the Bureau of Labor Statistics (BLS).
receive the full refundable amount or were ineligible for the benefit ATUS respondents are randomly selected from a subset of households
entirely. Indeed, families had to earn at least $2500 to claim the benefit. that have completed the American Current Population Survey (CPS)
The CTC phased in at a rate of 15% for each dollar above $2500, up survey and are only interviewed once. Using time diaries, the ATUS
to a maximum of $1400. This was the maximum refundable amount of asks respondents how they spent every minute of the preceding 24-h
the CTC, which could be supplemented with a non-refundable portion day (i.e., yesterday). The activities within a person’ s time diary are
of the tax credit, until reaching $2000. Thus, before the reform in 2021, classified into hundreds of detailed activity categories and include with
the CTC was not fully refundable. As a result, one in three children were whom the person spent their time. ATUS interviews take place every
ineligible for the full benefit because their parents did not earn enough day of the week (Monday through Sunday) to develop a representative
to be eligible. Children in single-parent families, in rural areas, from portrait of daily life; BLS interviewers choose the day of the interview.
ethnic and racial minorities, and in large families were most likely to The WB module is a supplementary survey included in the ATUS
be ineligible for the full credit (Collyer et al., 2019; Curran and Collyer, to obtain information on respondents’ emotional well-being. The WB
2020). module is available for three full years: 2010, 2012, and 2013. Another
In 2021, the ARP temporarily transformed the CTC into a near- collection of the WB module was added in the ATUS from March 1,
universal child benefit. Three fundamental changes took place: (i) the 2021, through December 31, 2021.
law extended eligibility to households with little or no income; (ii) the We use data from the 2021 ATUS and WB to estimate the 2021 CTC
law made the benefit fully refundable and distributed monthly for the expansion on well-being and time use. Data from 2010 to 2013 are used
second half of 2021; and (iii) the maximum credit amount increased for additional robustness checks at the end of the manuscript. In the
to an annual maximum of $3600 per child under age 6 and $3000 WB module, three activities reported by each respondent are randomly
per child ages 6 to 17. The first monthly payment for the expanded selected; for each of the selected activities, the respondent is asked how
CTC was made on July 15, 2021, and the final payment was made he/she felt during each of his/her activities. Five emotions are asked, in
on December 15, 2021. During this 6-month period, the credit was random order, for each of these three activities. This method is known
distributed monthly—namely, up to $300 per child under the age of as the day reconstruction method and is well established in existing
6 and up to $250 per child 6–17 years old. Thus, the reform changed evidence (Clark et al., 2020).
the frequency of benefits, with half of the annual amount as monthly We use the following outcomes: (1) happy; (2) stress; (3) pain; (4)
sad; and (5) tiredness. Each emotion is rated on a scale from 0 to 6.
A value of 0 denotes no emotion at all (for example, not happy at all)
1
For more information on the history of the CTC, see Garfinkel et al. (2016) whereas a value of 6 denotes the highest emotion (for example, very
and Crandall-Hollick (2018, 2021). happy). We also combine the negative emotions (stress, pain, sad, and

2
L. Lebihan Economics and Human Biology 59 (2025) 101533

Fig. 1. Child tax credit, before and after the 2021 expansion.
Notes: Shows child tax credits amounts by adjusted gross income (AGI), age of children, and marital status, before and after the 2021 expansion. The figure
illustrates the CTC credit amount as a function of household income before and after the 2021 expansion, for households with children aged 0–5 and/or 6–17 years.
The information is shown separately for single parents and married parents (panels a and b, respectively). See more details in the main text.

tiredness) to create a measure of “average negative affect.” We create receipt and amount of CTC benefits are correlated with other household
an overall measure of well-being, “net affect,” which is the difference characteristics that could be associated with well-being measures.
between the happiness score and average negative affect score. For Rather, here, following the existing literature, we use a simulated
each ATUS–WB survey respondent and for each of the three activities measure of household benefits based on policy changes (Currie and
(for which emotions were asked), we compute these seven measures Gruber, 1996). We calculate the monthly household CTC benefit as the
of well-being. We then take the average across the three activities. sum of $300 multiplied by the number of children under 6 years old
These measures are very common in the experience well-being litera- and $250 multiplied by the number of children 6 to 17 years old. For
ture (Doyle et al., 2017). In addition, we use self-reported health status respondents surveyed before July 15, the monthly CTC amount is $0,
as an outcome variable to measure an individual’s health with the regardless of the number and age of the children in the household.
question: “Would you say your health in general is . . . ? ” Responses Therefore, the variations in the model come from two sources: (1)
are coded on a 5-point Likert scale: 1 (excellent), 2 (very good), 3 between-household differences in the number and age of the children
(good), 4 (fair), and 5 (poor). Thus, for this health measure, higher (0–5 year old vs. 6–17 years) and (2) between-month differences (be-
values indicate worse health. We also construct two indicator variables:
fore and after CTC expansion). Our identification strategy is similar
a dummy that equals 1 if the individual is in excellent or very good
to the strategy employed by Pilkauskas et al. (2024b) and Pilkauskas
health and 0 otherwise, and a dummy that equals 1 if the individual
et al. (2024a) to understand the effects of CTC expansion on economic
reports fair or poor health and 0 otherwise. All these indicators are very
well-being. We estimate the following model:
common in the health economics literature (Lebihan, 2023). Details on
well-being measures are presented in Appendix B. 𝑌𝑖𝑡 = 𝛽0 + 𝛽1 𝐶𝑇 𝐶𝑖𝑡𝑐𝑎 + 𝛽2 𝑋𝑖𝑡 + 𝛿𝑠 + 𝛾𝑡 + 𝜃𝑐 + 𝜀𝑖𝑡 (1)
We also study the impact of the 2021 CTC expansion on parents’
time allocation. We analyze the following activities drawn from the where 𝑌𝑖𝑡 is an outcome variable for individual 𝑖 in month 𝑡. 𝐶𝑇 𝐶𝑖𝑡𝑐𝑎 is
ATUS time-use diary: (1) sleep; (2) paid work; (3) education; (4) home the predicted monthly benefit for a household according to the number
production; (5) leisure; (6) childcare; and (7) caregiving for adults. of children 𝑐 and the age of the children 𝑎. The parameter of interest,
Details on the ATUS time use diary are presented in Appendix C. 𝛽1 , represents the effect of a $100 increase in predicted CTC benefits on
For control variables, we use dummy variables for the sex of the well-being and time use. The term 𝑋𝑖𝑡 represents the control variables
respondent, age (18–29, 30–39, 40–49, 50–59, 60–64, 65 and older), listed in Section 3. Day fixed effects are also included. We include state
highest level of education (less than a high school diploma, high school fixed effects (𝛿𝑠 ) to control for time-invariant state-level characteristics
diploma, some college, bachelor’s degree or more), and race/ethnicity that influence individuals. Month fixed effects (𝛾𝑡 ) are also included in
(White, Black, Asian, Hispanic, other). Total household income is mea- the model. Because the number of children determines the CTC size and
sured in 7 intervals (less than $25,000; $25,000–$34,999; $35,000–$49,999; likely impacts both well-being and time use measures, we include fixed
$50,000–$74,999; $75,000–$99,999; $100,000–$149,999; effects for the number of children (𝜃𝑐 ). Finally, 𝜀𝑖𝑡 is the error term.
$150,000 or more).2 We use dummy variables for metropolitan areas This paper focuses on families with children. We restrict the sample
and the presence of a partner in the household. We also include number to respondents living with children under 18 years old in the household.
of children fixed effects. Some prior CTC research has used childless individuals as a comparison
group (Batra et al., 2023; Ananat et al., 2024; Pac and Berger, 2024).
4. Methodology However, as demonstrated in the next section, childless individuals
are likely to be a poor counterfactual in this study. Indeed, several
We use a DD framework to estimate the effect of the CTC on factors make it unlikely that the parallel trend assumption holds for
well-being and time use. ATUS–WB data do not contain information these groups. For these reasons, in this paper, the identification strategy
about the CTC receipt. Moreover, if we were able to use data about relies on variations in CTC benefits that parents receive according to
CTC receipt, we would have an endogeneity problem because the the number and age of the children residing in the household. To avoid
ambiguity, we also eliminate observations on the interview day, which
2
was July 15, 2021.
In ATUS–WB data, income is reported in 16 income ranges (less than
$5000; $5000–$7499; $7500–$9999; $10,000–$12,499; $12,500–$14,999;
In additional analyses, for the sake of completeness, we compare
$15,000–$19,999; $20,000–$24,999; $25,000–$29,999; $30,000–$34,999; households with and without children. Similar to the literature (Kovski
$35,000–$39,999; $40,000–$49,999; $50,000–$59,999; $60,000–$74,999; et al., 2023; Ananat et al., 2024), when we use childless households
$75,000–$99,999; $100,000–$149,999; $150,000 or more). as a comparison group, we measure the CTC in two ways. The first

3
L. Lebihan Economics and Human Biology 59 (2025) 101533

method includes a pre/post-indicator interacted with an indicator for Table 1
the presence of children in the household (binary measure): Effects of the 2021 CTC on well-being.
Variables CTC CTC CTC N
𝑌𝑖𝑡 = 𝛽0 + 𝛽1 𝐶ℎ𝑖𝑙𝑑017𝑖 + 𝛽2 𝑃 𝑜𝑠𝑡𝑡 + 𝛽3 (𝐶ℎ𝑖𝑙𝑑017 × 𝑃 𝑜𝑠𝑡)𝑖𝑡 + 𝛽4 𝑋𝑖𝑦 + 𝛿𝑠 (1) (2) (3)

+ 𝛾𝑡 + 𝜃𝑐 + 𝜀𝑖𝑡 (2) Happy 0.014 0.044 0.044 2040
(0.014) (0.033) (0.029)
where 𝐶ℎ𝑖𝑙𝑑017𝑖 is a binary variable equal to 1 if the respondent Stress 0.019 −0.006 −0.009 2044
(0.012) (0.022) (0.027)
has at least one child 0–17 years old living in the household and 0
Pain −0.016 −0.020 −0.008 2046
otherwise. The term 𝑃 𝑜𝑠𝑡𝑡 is a dummy variable that takes the value of (0.014) (0.018) (0.018)
1 if the observation occurred after July 15 (post-treatment period) and Sad −0.017* −0.044** −0.040** 2047
0 otherwise. Here, the parameter of interest is 𝛽3 , and represents the (0.009) (0.019) (0.019)
Tiredness 0.020 0.021 0.028 2045
effect of having a child 0–17 years old living in the household after
(0.018) (0.026) (0.031)
July 15. Negative affect 0.001 −0.010 −0.005 2036
The second method uses predicted CTC amounts (continuous mea- (0.010) (0.015) (0.018)
sure): Net affect 0.012 0.055 0.045 2029
(0.020) (0.039) (0.040)
𝑌𝑖𝑡 = 𝛽0 + 𝛽1 𝐶𝑇 𝐶𝑖𝑡𝑐𝑎 + 𝛽2 𝑋𝑖𝑡 + 𝛿𝑠 + 𝛾𝑡 + 𝜃𝑐 + 𝜀𝑖𝑡 (3) Health −0.030*** −0.055*** −0.049*** 2057
(0.009) (0.014) (0.012)
where 𝐶𝑇 𝐶𝑖𝑡𝑐𝑎 is the predicted monthly benefit for a household accord- Excellent/very good health 0.011* 0.020*** 0.018** 2057
ing to the number of children 𝑐 and the age of the children 𝑎. The (0.006) (0.007) (0.007)
Fair/poor health −0.012** −0.021*** −0.018*** 2057
monthly household CTC benefit is calculated as in the first equation,
(0.005) (0.007) (0.005)
with the additional information that, for households without children,
State fixed effects No Yes Yes
the CTC amount is always set to 0. Month fixed effects No Yes Yes
The ATUS–WB data do not provide information on actual CTC Day fixed effects No Yes Yes
receipt. Instead, we define treatment by CTC eligibility. The estimates Number of children fixed effects No Yes Yes
reported are intention-to-treat effects. Administrative records show that Individual X’s No No Yes

the temporarily expanded CTC covered 90% of families with children Notes: For each dependent variable, we report the estimated policy effects (𝛽1 of Eq.
in the U.S. (Ananat et al., 2024). This means that treatment effects are (1) shown). The coefficients represent the effect of an additional $100 in monthly
2021 CTC benefits. The sample is restricted to households with at least one child
unlikely to be driven by individuals not taking CTCs.
younger than 18. Three specifications are reported: (1) no control variables; (2) only
We use linear probability models for binary dependent variables and state, month, day, and number of children fixed effects; and (3) the addition of the
ordinary least squares (OLS) for continuous dependent variables. We individual control variables listed in Section 3 (i.e., gender, age, education, income,
also use the weights available in the ATUS–WB data. Standard errors metropolitan area, race/ethnicity, presence of a partner in the household). Standard
errors (in parentheses) are clustered at the state level.
are clustered at the state level.
* Significant at 10%.
Summary statistics for well-being and time use outcomes are shown
** Significant at 5%.
in Tables A1 and A2, respectively. We also present the summary *** Significant at 1%.
statistics for the independent variables in Table A3 of the Appendix.

5. Results a standard deviation).3 Estimates also showed a 1.8 percentage point
increase in the likelihood of being in excellent/very good health and
This section is arranged as follows. First, we report the estimates a 1.8 percentage point decrease in the likelihood of being in fair/poor
of CTC expansion on adults’ well-being and time use following our health. Taken together, these results suggest that the 2021 CTC reform
baseline specification that compares adults with children. Second, we may have improved the well-being and health of parents.
present additional results using an alternative identification strategy Table 2 presents results on the time spent on sleep, paid work
that compares adults with children to adults without children. Third, hours, education, home production, leisure, childcare, and caregiving
we present results from a series of robustness checks for our preferred for adults. The reform had no significant effect on any of these ac-
identification strategy. Fourth, we explore any heterogeneous effects. tivities, with the exception of childcare hours (Column 3). Indeed, an
additional $100 in monthly CTC benefits was associated with a 5%
5.1. Baseline results increase in childcare relative to the mean score of 1.38, which is 3% of
the standard deviation.4
Table 1 presents the estimated effects of the 2021 CTC expansion
on the well-being of adults with children. Three specifications are 5.2. Additional findings comparing households with and without children
reported: (i) no control variables; (ii) only state, month, day, and
number of children fixed effects; and (iii) the addition of the individual This paper focuses on households with children younger than 18
control variables listed in Section 3. In the three specifications, the years old, exploiting variation in the generosity of CTC benefits by
results are generally consistent. Thus, the remainder of this study the number and age of children in the households. We also present
focuses on the models with a full set of controls. Estimates suggest an alternative identification strategy that compares households with
that an additional $100 in monthly CTC benefits was associated with
a decrease in parental sadness among households with children. More
3
specifically, the reform produced a 0.040 decrease in sadness (i.e., a As indicated in the data section, responses related to the health status are
decrease of 7.05% relative to the mean score of 0.53), which is 3.21% coded on a 5-point Likert scale: 1 (excellent), 2 (very good), 3 (good), 4 (fair),
and 5 (poor). Higher values indicate worse health. Here, the reform produced
of a standard deviation. For the happiness, stress, pain, and tiredness
a decrease in health measure, so lower values indicate better health for the
measures, there was no evidence of a significant change following
parents following the CTC expansion.
the 2021 CTC reform. Similarly, the decrease in negative affect and 4
In Table A4 of the Appendix, we divide the childcare hours into several
the increase in net affect were not statistically significant. However, categories. We find a significant increase in time spent on caring for and
an additional $100 in monthly CTC benefits showed a beneficial and helping children. However, the reform had no significant impact on time spent
significant effect on the health status of individuals (about a 4.76% of for activities related to children’s education and health.

4
L. Lebihan Economics and Human Biology 59 (2025) 101533

Table 2 day as childless households (less leisure time, more time devoted
Effects of the 2021 CTC on time use. to childcare, etc.). To test this hypothesis, we present the trends of
Variables CTC CTC CTC N all outcomes for households with children versus households without
(1) (2) (3) children in Figures A1–A2. For all measures of well-being and time use,
Sleep −0.061*** −0.052* 0.010 2840 the trends between the two groups are not parallel during the pre-
(0.019) (0.027) (0.031)
reform months.6 Second, demographic characteristics differ between
Paid work 0.054 0.070 −0.032 2840
(0.040) (0.052) (0.058)
households with and without children. As Table A7 shows, households
Education −0.037*** −0.076** −0.031 2840 without children were more likely to be older, educated, and White
(0.011) (0.030) (0.025) and less likely to have a spouse. Third, other threats to the difference-
Home production 0.057* 0.036 0.018 2840 in-differences model are events or policy changes affecting one group
(0.029) (0.037) (0.038)
but not the other during the study period. During the COVID-19 period,
Leisure −0.122*** −0.150*** −0.039 2840
(0.021) (0.041) (0.048) the two groups were likely to react differently to shocks. For example,
Childcare 0.108*** 0.157*** 0.063*** 2840 parents may have faced more emotional difficulties in the fall of 2021
(0.017) (0.020) (0.022) related to the reopening of schools at that time. COVID-19 vaccines
Caregiving for adults −0.001 0.007 0.009 2840 were generally not yet available for children during this period, and the
(0.005) (0.007) (0.007)
risk of infection was therefore greater for unvaccinated children. Fur-
State fixed effects No Yes Yes thermore, school closures were still frequent during 2021, increasing
Month fixed effects No Yes Yes
stress for parents and disrupting their schedules. For all these reasons,
Day fixed effects No Yes Yes
Number of children fixed effects No Yes Yes our preferred strategy excludes households without children from the
Individual X’s No No Yes control group.
Notes: For each dependent variable, we report the estimated policy effects (𝛽1 of Eq.
(1) shown). The coefficients represent the effect of an additional $100 in monthly 5.3. Robustness checks
2021 CTC benefits. The sample is restricted to households with at least one child
younger than 18. Three specifications are reported: (1) no control variables; (2) only For the rest of the analysis, we return to our baseline specification,
state, month, day, and number of children fixed effects; and (3) the addition of the
where we focus only on households with children.
individual control variables listed in Section 3 (i.e., gender, age, education, income,
metropolitan area, race/ethnicity, presence of a partner in the household). Standard Our identification strategy relies on a combination of variations in
errors (in parentheses) are clustered at the state level. CTC benefits in terms of both the age and number of the children
* Significant at 10%. before and after the policy change. A key identification assumption of
** Significant at 5%. our model is the common time trend assumption. In order to examine
*** Significant at 1%. whether this is the case, we perform a series of event study models. In
Fig. 2, we report event study plots from regressions of our well-being
outcomes on all control variables7 as well as interactions of month in-
children under 18 (treated group) with households without children dicators with the number of child indicators (1, 2, or 3 or more). We do
under 18 (control group). the same for time use measures in Fig. 3. The figures plot the interaction
In Tables A5 and A6 of the Appendix, we present results comparing terms to test whether significant differences occur in outcome trends
households with and without children in terms of well-being and time among families of different sizes before the reform. The reference cate-
use measures, respectively. We measure the CTC in two ways: (i) with gory is parents with one child (omitted category). Gray lines represent
a pre/post-indicator interacted with an indicator for the presence of respondents with two children. Black lines represent respondents with
children in the household (binary measure) and (ii) with predicted three children or more. In Tables D1 and D2 (Appendix D), we report
CTC amounts (continuous measure). In the binary method, the effect estimates associated with Figs. 2 and 3, respectively. We show the
of the reform is modeled only by the presence of children in the interaction terms and test whether significant differences in outcome
household after the reform, whereas the continuous measure exploits trends emerge among families of different sizes before the reform.
variations in CTC amounts according to the number (including 0) and Using three specifications, we test whether all pre-reform coefficients
age of the children. When we use the binary measure, many of the are equal to 0. The first specification tests whether the interaction
effects on well-being and time use become non-significant. However, terms associated with respondents with two children (2child) before
when the continuous measure is used, we continue to find evidence of the reform are significantly different from 0. The second specification
improvements in sadness and health (Table A5). Similarly, we find a tests whether the interaction terms associated with respondents with
significant increase in childcare (Table A6). These results are hardly three or more children (3ormchild) before the reform are significantly
surprising using this alternative identification strategy. Previous stud- different from 0. The third specification tests whether all interaction
ies have reported that effects turn statistically non-significant when terms before the reform (regardless of the number of children) are
comparing households with and without children (Kovski et al., 2023; significantly different from 0. For the well-being measures, we show
Pignatti and Parolin, 2024).5 no significant differences (at the 5 percent level) in the months prior
We are reluctant to rely on models that use childless individuals to the policy change (Fig. 2 and Table D1). For the time use measures,
we generally find no significant differences (at the 5 percent level) in
as a comparison group because childless individuals are likely to be a
the months prior to the reform. Some exceptions are observed, such as
poor counterfactual in this study for several reasons. First, the crucial
sleep and childcare (Fig. 3 and Table D2).
hypothesis for DD methodology is the common trend assumption before
In Fig. 4, we report event study plots from regressions of our well-
the policy change. More precisely, the trends between the treatment
being outcomes on all control variables8 as well as interactions of
and control groups should be parallel before the reform. The parallel
trend assumption between the two groups is unlikely to hold during
the pre-reform period as parents are generally more stressed and tired 6
We also visually inspected trends for both groups by restricting the
before the policy compared to households without children. Households analysis to households aged 18–64 (Figures A3 and A4 for well-being and
with children also do not have the same uses of time during a typical time use, respectively). Again, we show that, for all measures of well-being
and time use, the pre-reform trends are not parallel between the two groups.
7
The control variables are gender, age, education, income, metropolitan
5
Existing evidence indicates that the lack of significance is more present area, race/ethnicity, presence of a partner in the household, state, and day
for the binary measure and, to a lesser extent, for the continuous measure. fixed effects.

5
L. Lebihan Economics and Human Biology 59 (2025) 101533

Fig. 2. Event study plots of parallel trends by number of children for well-being measures.
Notes: Event study plots from regressions of our well-being outcomes on all control variables (gender, age, education, income, metropolitan area, race/ethnicity,
presence of a partner in the household, state, and day fixed effects) as well as interactions of month indicators with number of child indicators (1, 2, or 3 or
more). The figures plot the interaction terms to test whether significant differences in well-being outcome trends emerge between families of different sizes before
the reform. Gray lines represent respondents with two children. Black lines represent respondents with three children or more. The reference category is parents
with one child (omitted category).

Fig. 3. Event study plots of parallel trends by number of children for time use measures.
Notes: Event study plots from regressions of our time use outcomes on all control variables (gender, age, education, income, metropolitan area, race/ethnicity,
presence of a partner in the household, state, and day fixed effects) as well as interactions of month indicators with number of child indicators (1, 2, or 3 or
more). The figures plot the interaction terms to test whether significant differences in time use outcome trends emerge between families of different sizes before
the reform. Gray lines represent respondents with two children. Black lines represent respondents with three children or more. The reference category is parents
with one child (omitted category).

6
L. Lebihan Economics and Human Biology 59 (2025) 101533

month indicators with the age of child indicators (children 0–5 only, model. We show that the results are similar to the original estimates.
children 0–5 and 6–17, or children 6–17 only). We do the same for We perform the same robustness checks for time use measures and
time use measures in Fig. 5. The figures plot the interaction terms to test show that our results are robust (Table A9). Additional CTC benefits are
whether significant differences occur in outcome trends among families significantly associated with an increase in childcare, but no significant
with children of different ages before the reform. The reference cate- effect is found for other time use measures.
gory is parents with children aged 6–17 only (omitted category). Gray In Tables A10 and A11 of the Appendix, we include 2010, 2012,
lines represent respondent parents with children aged 0–5 only. Black and 2013 in the pre-CTC comparison period according to different
lines represent parents with children aged 0–5 and 6–17. In Tables D3 specifications for well-being and time use measures, respectively.10 In
and D4 (Appendix D), we report estimates associated with Figs. 4 and 5, Panel A, we include 2010, 2012, and 2013 in the pre-CTC comparison.
respectively. We show the interaction terms to test whether significant In Panel B, we exclude the period from March 1, 2021, to July 14,
differences in outcome trends emerge among families of different ages 2021, and compare the months when the expansion took place in 2021
before the reform. Using three specifications, we test whether all pre- with the years 2010, 2012, and 2013. Indeed, the first payment took
reform coefficients are equal to 0. The first specification tests whether place on July 15, 2021, but the reform was enacted in March 2021. We
the interaction terms associated with respondents with children aged therefore want to rule out the hypothesis of household anticipation. In
0–5 only (05only) before the reform are significantly different from 0. Panel C, we limit the sample to July 16 through December 31 of each
The second specification tests whether the interaction terms associated year in order to compare the months when the CTC expansion took
with respondents with children aged 0–5 and 6–17 (05–617 ) before place in 2021 with the same months in 2010, 2012, and 2013. In Panels
the reform are significantly different from 0. The third specification D through L, we perform the same restrictions but detail and compare
tests whether all interaction terms before the reform (regardless of each year (2010, 2012, and 2013) one by one with the year 2021.11
the age of the children) are significantly different from 0. For the Regardless of the specification, we find that the results remain similar
well-being measures, we generally show no significant differences (at to the original estimates for well-being measures (Table A10). For time
the 5 percent level) in the months prior to the policy change (Fig. use measures, we continue to find evidence of an increase in childcare
4 and Table D3). The only exceptions are happy and net affect. For for all specifications. We also find a significant decrease in leisure time
the time use measures, some significant differences (at the 5 percent for some specifications (Table A11).
level) in the months prior to the reform are observed (education, home In Tables A12 and A13 of the Appendix, we allow the policy effects
production, leisure, and childcare; Fig. 5 and Table D4). We are aware to vary by month (July to December) for well-being and time use
that we must be cautious in our conclusions for these variables. As measures, respectively. We also test whether the effect is different
these results indicate, overall, there is little evidence of significant pre- from 0 as well as the equality of policy effects over time. For the
trends. However, we caution that these graphs do not perfectly capture majority of the well-being outcome (pain, sadness, net affect, negative
the variations we use in our models, as we use variations in both the affect, health, excellent/very good health, fair/poor health), the null
number and age of children in the household. hypothesis of the equality of effects between the months is rejected,
suggesting that the effects differ over time (Table A12). More generally,
In Table A8, we present estimates from a number of robustness
by observing each variable, we find that the beneficial effects of the
checks for well-being measures. Each panel represents a different re-
reform appeared mainly from the month of August (i.e., when more
gression. The original estimates are shown in the first row. Panel
households are affected) and persisted until December. Turning to the
A excludes the first month of CTC payment (July). Indeed, existing
time use measures, we report that the reform increased childcare from
evidence shows that families are less likely to receive the benefit in the
the month of August and the effects persisted until December (Table
first month of the reform (Kovski et al., 2023). Excluding July, the re-
A13). We find no significant effects for the other time use outcomes.
sults are even more compelling. A significant decrease in pain, sadness,
However, the sample size may be too small to detect any significance
and the likelihood of being in fair/poor health occurs. Additional CTC
of the effects by month.
benefits are also associated with a significant improvement in health
In Table A14 of the Appendix, we use ordered probit regressions for
and an increase in the likelihood of being in excellent/very good health.
ordered variables (happiness, stress, pain, sadness, tiredness, net affect,
Panel B excludes the last month of the observation window (December)
negative affect, health) and probit marginal effects for binary variables
to test whether families react differently when payments are about
(excellent/very good health, fair/poor health). We show that our results
to end. Although the effects on emotions become non-significant, the
are robust to these specifications.
beneficial effects on health remain significant. In Panel C, we exclude
Tables A15 and A16 of the Appendix restrict the sample to house-
households with children aged 11–17 in order to account for the fact
holds aged 18–64 for well-being and time use measures, respectively.
that the COVID-19 vaccine became available in the fall of 2021. The
We show that the results are similar to our main estimates.
results are similar when excluding them. We also find a significant
According to Kling et al. (2007), aggregating multiple measures into
increase in happiness and net affect. Panel D adds employment status
an index in a particular domain (e.g., emotional well-being) improves
as a covariate into the model and shows that the results are similar to
statistical power. So, following Hoynes et al. (2016), we estimate two
the original estimates.9 Panel E adds as a control an indicator for the different summary standardized indexes of emotional well-being. The
presence of any children aged 0–5 in the household and shows that the first index includes four measures: stress, pain, sadness, and tiredness.
results remain similar to the original estimates. In Panel F, we include An increase in the first index indicates a worse outcome. The second
fixed effects for the age of the youngest child in the household in the include five measures: happiness, stress, pain, sadness, and tiredness.
We convert the happiness variable so that an increase of the second

8
The control variables are gender, age, education, income, metropolitan
10
area, race/ethnicity, presence of a partner in the household, state, and day As previously mentioned, the WB module is available for three full years:
fixed effects. 2010, 2012, and 2013. Another collection of the WB module was included in
9
We do not control for employment status in the main estimates because the ATUS from March 1, 2021, through December 31, 2021.
11
the 2021 CTC reform could have had effects on labor market outcomes. Later, For time use measures, we add a specification that excluded January
we test this and show that the reform had no significant impact on paid and February 2021 (Panel M of Table A11). Indeed, well-being measures are
work activities. The lack of effect on employment is consistent with existing available for the period from March 2021 to December 2021, while time
studies (Ananat et al., 2024; Pac and Berger, 2024). The fact that the effects on measures are available from January 2021 to December 2021. We show that,
well-being and health remain similar when controlling for employment status by restricting the analysis to the March 2021–December 2021 period for time
suggests that other transmission channels must be at play. use measures, the results remain identical.

7
L. Lebihan Economics and Human Biology 59 (2025) 101533

Fig. 4. Event study plots of parallel trends by age of children for well-being measures.
Notes: Event study plots from regressions of our well-being outcomes on all control variables (gender, age, education, income, metropolitan area, race/ethnicity,
presence of a partner in the household, state, and day fixed effects) as well as interactions of month indicators with age of child indicators in the household
(children aged 0–5 only, children aged 0–5 and children aged 6–17, and children aged 6–17 only). The figures plot the interaction terms to test whether significant
differences in well-being outcome trends emerge between families with children from different age categories before the reform. Gray lines represent respondents
with only children aged 0–5 in the household. Black lines represent respondents with children aged 0–5 and 6–17 in the household. The reference category is
parents with only children aged 6–17 in the household (omitted category).

index also indicates a worse outcome. The summary index is the simple Estimates show that the effects on well-being measures are similar,
average of the standardized z-scores measured for each component. regardless of education. In Panel C, we study how effects vary across
The z-score is calculated by subtracting the mean and dividing by the income. Low-income respondents include individuals with an annual
standard deviation.12 Table A17 of the Appendix presents the results for family income less than $50,000. Estimates show that poor individuals
the two indexes. For the first index, the coefficient estimate is similar are less tired after the policy change. The beneficial effect of the reform
to that obtained for the negative affect outcome in Table 1, and there on health is also larger for low-income respondents. In Panels D and
is no evidence of a significant change following the 2021 CTC reform. E, the policy effects on well-being are generally similar regardless of
For the second index (including happiness), we show that the effects the race and marital status. In Panel F, we test how the policy effects
are also insignificant.13 vary with children of different ages. In particular, we want to know
whether the effects differ depending on whether the household has at
5.4. Heterogeneous effects least one child aged 0–5 or not. Indeed, childcare needs and time vary
depending on the child’s age. For well-being measures, the effects are
In Table 3, we evaluate the heterogeneous effects of the policy similar regardless of the presence or not of at least one child aged 0–5
for well-being measures. We study how effects vary across gender, in the household.14
education, income, race, marital status, and child’s age. In the base- Table 4 presents the heterogeneous effects of the reform on time
line specification (Eq. (1)), we add an interaction term between the use measures. The results in Panel A reveal that the effects are similar
subgroup and the predicted benefit. Results in Panel A reveal that the regardless of the respondent’s sex. The only exceptions are education
effects are similar regardless of the respondent’s sex. The only exception and leisure, where we observe a significant decrease in education
is sadness, where we observe that the effects are only significant for and an increase in leisure for women. In Panel B, estimates report a
women. In Panel B, we study how effects vary across education. Low- significant increase in paid work and a significant decrease in leisure
for low-educated respondents. In Panel C, we study how effects vary
educated individuals include those with a high school diploma or less.
across income. We found that, for low-income individuals, time in home
production increased while time in childcare decreased. The results in
12
Like Hoynes et al. (2016), we use the mean and standard deviation of the
Panel D show that the policy effects on well-being are generally similar
pre-reform period. regardless of the race, except for sleep. In Panel E, estimates show a
13
It is normal that the estimated coefficient for the second index in Table significant increase in paid work and a significant decrease in leisure for
A17 is different from that estimated for the net affect variable in Table 1. In single mothers. In Panel F, we report a significant increase in childcare
Table A17, the happiness variable has been converted to unhappiness, where
a higher score is a worse outcome. This is to be consistent with the other
14
negative emotions (stress, pain, sadness, tiredness). In Table 1, the net affect For Panel F, we also include in the model an indicator for any children
variable is the difference between the (untransformed) happiness score and aged 0–5 in the household. If this variable is not included, the significance of
the average negative affect score (negative emotions). the effects obtained is larger than without it.

8
L. Lebihan Economics and Human Biology 59 (2025) 101533

Fig. 5. Event study plots of parallel trends by age of children for time use measures.
Notes: Event study plots from regressions of our time use outcomes on all control variables (gender, age, education, income, metropolitan area, race/ethnicity,
presence of a partner in the household, state, and day fixed effects) as well as interactions of month indicators with age of child indicators in the household
(children aged 0–5 only, children aged 0–5 and children aged 6–17, and children aged 6–17 only). The figures plot the interaction terms to test whether significant
differences in well-being outcome trends emerge between families with children from different age categories before the reform. Gray lines represent respondents
with only children aged 0–5 in the household. Black lines represent respondents with children aged 0–5 and 6–17 in the household. The reference category is
parents with only children aged 6–17 in the household (omitted category).

only for parents with at least one child aged 0–5 in the household. in the opposite direction as expected (i.e., emotions got worse). Only for
Clearly, there is a significant difference between younger children and happiness do we find a beneficial effect, but the effect is only significant
older children.15 at 10% (Table A18).
As previously mentioned, the CTC benefits begin to phase out at Turning to the time use measures (Table A19), the results are similar
$112,500 for single parents and $150,000 for married parents. In Ta- to the original estimates. Families with incomes below $150,000 are
bles A18 and A19, we report several alternative samples for well-being the ones most affected by the policy change and are seeking a signif-
and time measures, respectively. In Panel A, we exclude respondents icant increase in childcare. In contrast, for families with high income
whose household income is $150,000 or more.16 This approach allows ($150,000 or more), there is no significant change (at a significance
us to focus exclusively on families that were most heavily impacted level of 5% or less) in the time use measures.
by the 2021 CTC reform (based on income distribution, the credit size
increased most for low-income families) and whose health and well- 6. Discussion and conclusion
being were most negatively impacted by the pandemic. The results
are similar to the original estimates. Thus, we observe a significant This paper presents new empirical evidence on the effects of the
decrease in sadness and in the likelihood of being in poor/fair health 2021 CTC expansion on well-being and time-use activities of par-
(Table A18). Estimates also show that health improved and the likeli- ents. This paper shows that the reform decreased parents’ sadness and
hood of being in excellent/very good health increased. Panel B restricts improved their health. In Table A20, we present a summary table
the sample to households with a family income of $150,000 or more, comparing our results with related literature on the CTC expansion’s
which can be considered a placebo group because these households effects on parental health. The beneficial effects of the policy on overall
were not affected by the reform or, even if they experienced an increase health, as found in this paper, are consistent with existing evidence.
in benefits, the effect should be smaller compared to poorer households. For example, according to Rook et al. (2023), CTC-eligible adults
Reassuringly, we find no significant impact of the reform on health in experienced a 3.0 percentage point increase in the likelihood of being in
this group. Some significant effects are observed on pain, but they are excellent/very good health—an estimate close to ours. Several studies
also show that CTC expansion had a beneficial effect on parental mental
health. Specifically, existing evidence reports a decrease depressive
15
For Panel F, we also include in the model an indicator for any children
and anxiety symptoms (Batra et al., 2023; Cha et al., 2023; Kovski
aged 0–5 in the household. If this variable is not included, the significance
et al., 2023; Pignatti and Parolin, 2024). Our data do not include
and the size of the effects obtained are larger than without it.
16
In the ATUS–WB data, income is reported in income ranges (less
information on depressive and anxiety symptoms, but we find that the
than $25,000, $25,000-$34,999, $35,000-$49,999, $50,000-$74,999, $75,000- CTC expansion led to a significant decrease in sadness for parents.
$99,999, $100,000-$149,999, $150,000 or more). We cannot distinguish the Sadness is considered one of the key symptoms of depression and is
phase-out threshold of $112,500 for single parents. Instead, we use the same strongly correlated with anxiety and depression (Mouchet-Mages and
criterion of $150,000 for single parents and married parents. Bayle, 2008; Zaid et al., 2025). Both mental health and physical health

9
L. Lebihan Economics and Human Biology 59 (2025) 101533

Table 3
Heterogeneous effects on well-being.
Variables Happy Stress Pain Sad Tiredness Negative affect Net affect Health Excellent/very Fair/poor
good health health
Panel A: By sex
CTC*Female −0.008 0.005 0.002 −0.030** 0.027 0.002 −0.004 0.015 0.003 0.015**
(0.029) (0.045) (0.029) (0.013) (0.035) (0.020) (0.039) (0.014) (0.008) (0.007)
CTC 0.048 −0.011 −0.009 −0.024 0.012 −0.006 0.048 −0.057*** 0.016* −0.027***
(0.033) (0.047) (0.026) (0.020) (0.040) (0.025) (0.048) (0.013) (0.009) (0.006)
Panel B: By education
CTC*LowEduc −0.017 0.047 0.005 0.001 0.026 0.020 −0.040 −0.014 −0.006 −0.008
(0.026) (0.031) (0.031) (0.014) (0.033) (0.021) (0.035) (0.020) (0.010) (0.008)
CTC 0.055* −0.041 −0.011 −0.041** 0.010 −0.019 0.073** −0.039* 0.022** −0.013**
(0.029) (0.032) (0.028) (0.020) (0.032) (0.018) (0.035) (0.021) (0.011) (0.006)
Panel C: By income
CTC*Less50k −0.005 −0.000 −0.018 −0.024 −0.103** −0.036 0.029 −0.039* 0.011 −0.003
(0.032) (0.030) (0.041) (0.021) (0.046) (0.028) (0.051) (0.021) (0.013) (0.008)
CTC 0.045 −0.009 −0.000 −0.030* 0.071** 0.011 0.033 −0.032** 0.014 −0.017***
(0.029) (0.027) (0.025) (0.017) (0.028) (0.019) (0.037) (0.015) (0.009) (0.006)
Panel D: By race
CTC*White 0.010 −0.050 −0.067 −0.027 −0.054 −0.051 0.063 −0.011 0.005 0.010
(0.042) (0.044) (0.051) (0.030) (0.059) (0.038) (0.070) (0.018) (0.010) (0.007)
CTC 0.035 0.034 0.049 −0.018 0.074 0.038 −0.008 −0.040* 0.014 −0.026***
(0.047) (0.053) (0.053) (0.030) (0.065) (0.039) (0.076) (0.020) (0.011) (0.007)
Panel E: By marital status
CTC*Single 0.009 0.079* 0.071 0.036 −0.026 0.041 −0.032 −0.006 −0.003 0.005
(0.053) (0.042) (0.047) (0.036) (0.046) (0.035) (0.074) (0.031) (0.014) (0.009)
CTC 0.042 −0.021 −0.019 −0.046** 0.032 −0.011 0.051 −0.048*** 0.019** −0.019***
(0.029) (0.027) (0.017) (0.018) (0.029) (0.017) (0.038) (0.012) (0.007) (0.005)
Panel F: By age of the child
CTC*Anychild05 0.022 0.056 0.034 0.008 0.035 0.035 −0.009 0.031* −0.017 −0.000
(0.028) (0.036) (0.034) (0.023) (0.031) (0.022) (0.040) (0.018) (0.011) (0.007)
CTC 0.029 −0.048 −0.030 −0.045* 0.005 −0.029 0.052 −0.069*** 0.030*** −0.018***
(0.035) (0.038) (0.032) (0.024) (0.038) (0.025) (0.050) (0.017) (0.010) (0.007)

Notes: Shows the heterogeneous effects of the policy for well-being measures. Each panel represents a different regression (by sex, education, income, race, marital status, and
child’s age). In the baseline specification (Eq. (1)), an interaction term between the subgroup and the predicted benefit is added. The coefficients represent the effect of an additional
$100 in monthly 2021 CTC benefits. The sample is restricted to households with at least one child younger than 18. All models include state, month, day, and number of children
fixed effects. They also include the individual control variables listed in Section 3 (gender, age, education, income, metropolitan area, race/ethnicity, and presence of a partner in
the household). Low-educated respondents include those with a high school diploma or less. For Panel F, we also include in the model an indicator for any children aged 0–5 in
the household. Standard errors (in parentheses) are clustered at the state level.
* Significant at 10%.
** Significant at 5%.
*** Significant at 1%.

are correlated with measures of subjective well-being (Dolan et al., mothers. Another study by Gangopadhyaya et al. (2020) found that a
2008). Our paper therefore shows that the reform has an impact on $1000 increase in maximum EITC payments decreased the number of
not only health, but also subjective well-being. Clearly, the effects are days not in good mental health by 0.21–0.27 days. Other cash transfers
consistent with prior research on CTC and health/well-being. programs have also been shown to increase money and time spent on
We also estimated the effect of the reform on time use activities. and with infants (Gennetian et al., 2022). By exploiting child tax benefit
Our estimates show that the policy had generally no significant impact expansion in Canada, Milligan and Stabile (2011) found positive effects
on work time. These results are consistent with existing evidence on the on parents’ mental health. Although the design of the programs differs
effects of CTC expansion on employment outcomes (Ananat et al., 2024; from the expanded CTC, all of these studies show the beneficial effect
Pac and Berger, 2024). Similarly, we do not find that the reform had of child benefits on parents’ well-being, which is consistent with the
a significant effect on time spent on health care, which is consistent results of our paper.17
with previous evidence of the null effects of the reform on health The heterogeneous effects show that the effects of the 2021 CTC
care use (Pignatti and Parolin, 2024). Therefore, we can exclude the expansion are larger for low-income households in terms of well-being
possibility that the beneficial effects on parental well-being were due and health. This is consistent with existing evidence on CTC expansion
to changes in employment or changes in health care use. Rather, here, it and more generally with child benefits (Pignatti and Parolin, 2024).
is hypothesized that the introduction of unconditional supplementary These people are most vulnerable to financial constraints and more
income reduced poverty and alleviated financial worries, which was likely to have poorer health as well as high marginal returns of benefits.
beneficial for parental health and well-being (Rook et al., 2023; Pignatti This result has important implications for policymakers because it
and Parolin, 2024). An additional hypothesis would be that parents shows that, even if it is temporary and during a particular period
spending more time with their children would be beneficial for the (COVID-19), receiving extra income is beneficial to the health of the
parents’ well-being (Musick et al., 2016). most vulnerable. This type of policy could lead to a reduction in health
Our findings are consistent with prior evidence on cash transfer disparities. Our findings also show that the effects are not significantly
programs, such as the Earned Income Tax Credit (EITC), which has higher for single parents or low-educated households. This result might
been associated with improvements in parental well-being and health.
For example, Boyd-Swan et al. (2016) showed that eligible mothers
experienced an increase in happiness and self-esteem as well as a de- 17
The EITC is conditioned on work. The Baby’s First Year is a large-
crease in depression symptomatology. Similarly, Evans and Garthwaite scale multi-site randomized controlled trial where the new mothers received
(2014) reported improvements in the self-reported health of eligible monthly unconditional cash transfers for several years (Gennetian et al., 2022).

10
L. Lebihan Economics and Human Biology 59 (2025) 101533

Table 4
Heterogeneous effects on time use.
Variables Sleep Paid work Education Home production Leisure Childcare Caregiving
for adults
Panel A: By sex
CTC*Female −0.043 −0.062 −0.029* −0.052 0.160*** 0.031 −0.006
(0.032) (0.076) (0.015) (0.049) (0.043) (0.033) (0.011)
CTC 0.036 0.004 −0.014 0.049 −0.133** 0.044** 0.012
(0.042) (0.084) (0.021) (0.049) (0.062) (0.022) (0.012)
Panel B: By education
CTC*LowEduc −0.015 0.156* −0.007 0.059 −0.162*** −0.018 −0.001
(0.041) (0.083) (0.014) (0.054) (0.045) (0.035) (0.008)
CTC 0.019 −0.134* −0.026 −0.020 0.066 0.075** 0.010*
(0.023) (0.073) (0.026) (0.043) (0.058) (0.032) (0.005)
Panel C: By income
CTC*Less50k 0.035 −0.083 0.008 0.089* −0.011 −0.056* 0.015
(0.034) (0.088) (0.020) (0.053) (0.049) (0.028) (0.013)
CTC −0.005 0.004 −0.034 −0.020 −0.035 0.087*** 0.002
(0.029) (0.063) (0.025) (0.040) (0.053) (0.029) (0.005)
Panel D: By race
CTC*White 0.114** −0.154 −0.012 0.067 −0.008 −0.015 0.019
(0.044) (0.107) (0.030) (0.070) (0.053) (0.044) (0.011)
CTC −0.084* 0.094 −0.021 −0.036 −0.033 0.075 −0.007
(0.048) (0.121) (0.040) (0.058) (0.073) (0.046) (0.010)
Panel E: By marital status
CTC*Single −0.020 0.234** −0.034 −0.027 −0.191*** 0.022 −0.001
(0.050) (0.103) (0.045) (0.055) (0.054) (0.037) (0.011)
CTC 0.013 −0.068 −0.026 0.022 −0.011 0.059** 0.009
(0.030) (0.059) (0.023) (0.039) (0.050) (0.023) (0.008)
Panel F: By age of the child
CTC*Anychild05 −0.005 0.013 −0.004 −0.034 −0.036 0.059* 0.003
(0.044) (0.067) (0.019) (0.043) (0.045) (0.033) (0.010)
CTC 0.013 −0.049 −0.028 0.042 −0.017 0.032 0.007
(0.049) (0.075) (0.030) (0.052) (0.058) (0.029) (0.008)

Notes: Shows the heterogeneous effects of the policy for time use measures. Each panel represents a different regression (by sex, education,
income, race, marital status, and child’s age). In the baseline specification (Eq. (1)), an interaction term between the subgroup and the predicted
benefit is added. The coefficients represent the effect of an additional $100 in monthly 2021 CTC benefits. The sample is restricted to households
with at least one child younger than 18. All models include state, month, day, and number of children fixed effects. They also include the
individual control variables listed in Section 3 (gender, age, education, income, metropolitan area, race/ethnicity, and presence of a partner
in the household). Low-educated respondents include those with a high school diploma or less. For Panel F, we also include in the model an
indicator for any children aged 0–5 in the household. Standard errors (in parentheses) are clustered at the state level.
* Significant at 10%.
** Significant at 5%.
*** Significant at 1%.

be surprising because it is not consistent with those of Pignatti and the ability to draw broader conclusions about the effect of monthly
Parolin (2024) and Cha et al. (2023). Indeed, these authors report a universal cash transfers on well-being and time use. Indeed, this reform
larger beneficial effect on mental health for more credit-constrained was temporary and lasted only six months. This program was also
individuals (including low-educated households, low-income families, implemented during a particular period—namely, the COVID-19 pan-
and single families). Several reasons could explain this discrepancy.
demic. During this period, sharp increases in unemployment rates and
First, our sample size is small (approximately 2040 observations for
well-being measures in the overall sample). Standard errors increase declining incomes, as well as a deterioration in people’s mental health
mechanically when performing subgroup analyses, thereby limiting our and well-being, were found. Thus, the effects of the CTC reform may
ability to draw strong conclusions from these results. Second, there have been particularly beneficial during this atypical period, yet it may
could be differences in the measurements. Indeed, Pignatti and Parolin limit the generalizability of results. Some seemingly eligible families
(2024) focused on the number of bad mental health days and Cha did not receive the benefits, and some payments were delayed (Pilka-
et al. (2023) on psychological distress symptoms. In our study, these are uskas et al., 2024b). Nevertheless, Michelmore and Pilkauskas (2023)
measures of general well-being and health (health, happiness, sadness, showed that more than 80% of low-income families received at least
etc.). As the measures can be very broad and include both mental and
one payment. The data used in this paper ended in 2021, so future
physical well-being, the effects obtained may be different. Nevertheless,
research should examine whether the effects of the CTC reform persist.
we find that the beneficial effects of the reform are larger for low-
income families. Income here appears to be a better indicator of family Although this is less of a concern given that the expanded CTC was tem-
vulnerability, especially as the CTC benefit eligibility criterion is based porary, it is difficult to generalize our results to a permanent expansion,
on income (before and after the expansion). All of our results by and individuals may respond differently to a more permanent change in
income18 are consistent with the fact that the families most likely to terms of employment, behavior, and health. It would also be interesting
be affected by the reform are those who see the largest effects on to see the effects when the program is returned to its original pre-2021
well-being and time use (households lower than $112,500 for single form. Finally, the sample sizes are small, making it difficult to measure
households and $150,000 for married households).
some effects. Nevertheless, the results found in this paper are consistent
Considering all the results, this paper shows that the 2021 CTC re-
with the existing literature on the CTC expansion. These findings have
form improved parents’ well-being and health, but several factors limit
important implications for policymakers and contribute to the ongoing
debate on the introduction of a permanent universal child benefit for
18
See Tables 3, 4, A18, and A19. the U.S. population.

11
L. Lebihan Economics and Human Biology 59 (2025) 101533

CRediT authorship contribution statement Doyle, O., Delaney, L., OFarrelly, C., Fitzpatrick, N., Daly, M., 2017. Can early
intervention improve maternal well-being? Evidence from a randomized controlled
trial. PLoS One 12 (1), e0169829.
Laetitia Lebihan: Writing – review & editing, Writing – origi-
Ettman, C.K., Abdalla, S.M., Cohen, G.H., Sampson, L., Vivier, P.M., Galea, S., 2020.
nal draft, Visualization, Validation, Supervision, Software, Resources, Prevalence of depression symptoms in US adults before and during the COVID-19
Project administration, Methodology, Investigation, Funding acquisi- pandemic. JAMA Netw. Open 3 (9), e2019686.
tion, Formal analysis, Data curation, Conceptualization. Evans, W.N., Garthwaite, C.L., 2014. Giving mom a break: The impact of higher EITC
payments on maternal health. Am. Econ. J. Econ. Policy 6 (2), 258–290.
Gangopadhyaya, A., Blavin, F., Braga, B., Gates, J., 2020. Credit where it is due:
Declaration of competing interest Investigating pathways from earned income tax credit expansion to maternal mental
health. Heal. Econ. 29 (9), 975–991.
The authors declare that they have no known competing finan- Garfinkel, I., Harris, D., Waldfogel, J., Wimer, C., 2016. Doing more for our children:
cial interests or personal relationships that could have appeared to Modeling a universal child allowance or more generous child tax credit, Vol.
22, New York: The Century Foundation, Bernard L. Schwartz Rediscovering
influence the work reported in this paper.
Government Initiative, (Accessed 22 September 2017).
Gennetian, L.A., Duncan, G., Fox, N.A., Magnuson, K., Halpern-Meekin, S., Noble, K.G.,
Appendix A. Supplementary data Yoshikawa, H., 2022. Unconditional cash and family investments in infants:
Evidence from a large-scale cash transfer experiment in the US. Natl. Bur. Econ.
Supplementary material related to this article can be found online Res..
Hoynes, H., Schanzenbach, D.W., Almond, D., 2016. Long-run impacts of childhood
at https://doi.org/10.1016/j.ehb.2025.101533. access to the safety net. Am. Econ. Rev. 106 (4), 903–934.
Kling, J.R., Liebman, J.B., Katz, L.F., 2007. Experimental analysis of neighborhood
Data availability effects. Econom. 75 (1), 83–119.
Kovski, N., Pilkauskas, N.V., Michelmore, K., Shaefer, H.L., 2023. Unconditional cash
transfers and mental health symptoms among parents with low incomes: Evidence
Data will be made available on request.
from the 2021 child tax credit. SSM Popul. Heal. 22, 101420.
Lebihan, L., 2023. The impact of a mandatory universal drug insurance program on
health behaviors and outcomes. Heal. Econ. 32 (9), 2006–2046.
References Michelmore, K., Pilkauskas, N.V., 2023. The 2021 child tax credit: Who received it and
how did they spend it? AEA Pap. Proc. 113, 413–419.
Ananat, E., Glasner, B., Hamilton, C., Parolin, Z., Pignatti, C., 2024. Effects of the Milligan, K., Stabile, M., 2011. Do child tax benefits affect the well-being of children?
expanded child tax credit on employment outcomes. J. Public Econ. 238 (105168). Evidence from Canadian child benefit expansions. Am. Econ. J.: Econ. Policy 3 (3),
Batra, A., Jackson, K., Hamad, R., 2023. Effects of the 2021 expanded child tax credit 175–205.
on adults mental health: A quasi-experimental study. Health Aff. 42 (1), 74–82. Mouchet-Mages, S., Bayle, F.J., 2008. Sadness as an integral part of depression.
Boyd-Swan, C., Herbst, C.M., Ifcher, J., Zarghamee, H., 2016. The earned income tax Dialogues Clin. Neurosci. 10 (3), 321–327.
credit, mental health, and happiness. J. Econ. Behav. Organ. 126, 18–38. Musick, K., Meier, A., Flood, S., 2016. How parents fare: Mothers and fathers subjective
Cha, E., Lee, J., Tao, S., 2023. Impact of the expanded child tax credit and its expiration well-being in time with children. Am. Sociol. Rev. 81 (5), 1069–1095.
on adult psychological well-being. Soc. Sci. Med. 332, 116101. Pac, J., Berger, L.M., 2024. Quasi-experimental evidence on the employment effects of
Clark, A.E., Doyle, O., Stancanelli, E., 2020. The impact of terrorism on individual the 2021 fully refundable monthly child tax credit. J. Policy Anal. Manag. 43 (1),
well-being: Evidence from the Boston Marathon bombing. Econ. J. 130 (631), 192–213.
2065–2104. Pignatti, C., Parolin, Z., 2024. The effects of an unconditional cash transfer on parents’
Collyer, S., Harris, D., Wimer, C., 2019. Left behind: The one-third of children in mental health in the United States. Heal. Econ..
families who earn too little to get the full child tax credit. Poverty Soc. Policy Pilkauskas, N., Michelmore, K., Kovski, N., 2024b. The effects of the 2021 child tax
Brief 3 (6), 1–6. credit on housing affordability and the living arrangements of families with low
Crandall-Hollick, M., 2018. The Child Tax Credit: Current Law. CRS Report R41873. incomes. Demogr. 61 (4), 1069–1096.
Washington, DC Congressional Research Service. Pilkauskas, N., Michelmore, K., Kovski, N., Shaefer, H.L., 2024a. The expanded child
Crandall-Hollick, M., 2021. The Child Tax Credit: Proposed Expansion in the American tax credit and economic wellbeing of low-income families. J. Popul. Econ. 37 (4),
Rescue Plan Act of 2021 (ARPA; HR 1319). Congressional Research Service. 68.
Curran, M., 2022. Research roundup of the expanded child tax credit: One year on. Rook, J.M., Yama, C.L., Schickedanz, A.B., Feuerbach, A.M., Lee, S.L., Wisk, L.E., 2023.
Poverty Soc. Policy Rep. 6 (9), 1–45. Changes in self-reported adult health and household food security with the 2021
Curran, M., Collyer, S., 2020. Children left behind in larger families: The uneven receipt expanded Child Tax Credit monthly payments. JAMA Heal. Forum 4 (6), e231672.
of the federal child tax credit by childrens family size. Poverty Soc. Policy Brief 4 Schild, J., Collyer, S.M., Garner, T., Kaushal, N., Lee, J., Waldfogel, J., Wimer, C.T.,
(4). 2023. Effects of the expanded child tax credit on household spending: Estimates
Currie, J., Gruber, J., 1996. Health insurance eligibility, utilization of medical care, based on US consumer expenditure survey data. Natl. Bur. Econ. Res..
and child health. Q. J. Econ. 111 (2), 431–466. Vindegaard, N., Benros, M.E., 2020. COVID-19 pandemic and mental health conse-
Department of Treasury, 2021. Treasury and IRS announce families of nearly 60 million quences: Systematic review of the current evidence. Brain Behav. Immun. 89,
children receive 15 billion in first payments of expanded and newly advanceable 531–542.
child tax credit. Zaid, S.M., Hutagalung, F.D., Abd Hamid, H.S.B., Taresh, S.M., 2025. The power of
Dolan, P., Peasgood, T., White, M., 2008. Do we really know what makes us happy? emotion regulation: how managing sadness influences depression and anxiety? BMC
A review of the economic literature on the factors associated with subjective Psychol. 13 (1), 38.
well-being. J. Econ. Psychol. 29 (1), 94–122.

File and source

File: effects-of-the-2021-expanded-child-tax-credit-on-parents-well-being-and-time-use.pdf
Size: 2,002,634 bytes
SHA-256: 0f5bf7e6b73a19db85a989cbb3ce0a86ec6787cc1554241a6d640b2828df58c8
Our copy: effects-of-the-2021-expanded-child-tax-credit-on-parents-well-being-and-time-use.pdf
Original: doi.org